1 00:00:00,000 --> 00:00:01,434 When you're really small, everyone understands what 2 00:00:01,434 --> 00:00:03,972 they're supposed to do. And at some point you'll cross a 3 00:00:03,972 --> 00:00:05,940 threshold where nobody does the whole code base. 4 00:00:05,940 --> 00:00:08,520 Nobody understands truly what's going on across the business. 5 00:00:08,910 --> 00:00:12,034 Michi Kono, CTO of Garner Health with experience leading teams at 6 00:00:12,034 --> 00:00:15,504 Meta and Stripe, reveals the art and science of scaling 7 00:00:15,504 --> 00:00:18,345 hypergrowth engineering teams. Management systems are about how 8 00:00:18,345 --> 00:00:22,140 you can run a group of people. It's a set of processes and 9 00:00:22,140 --> 00:00:24,666 expectations that can scale. It's about scaling repeatable 10 00:00:24,666 --> 00:00:26,956 outcomes. If you don't formalize it, it's 11 00:00:26,956 --> 00:00:29,190 basically tribal. You wanna make sure that as you 12 00:00:29,190 --> 00:00:32,040 scale, you codify these things. What's your tips and strategy to 13 00:00:32,040 --> 00:00:33,335 nurture a good engineering culture? 14 00:00:33,395 --> 00:00:36,100 I think a good engineering culture is one that focuses on 15 00:00:36,100 --> 00:00:38,484 engineering. You want a world where your best 16 00:00:38,484 --> 00:00:40,205 engineers are engineering the right systems. 17 00:00:40,295 --> 00:00:42,545 The only counterbalance is really the business outcomes. 18 00:00:42,575 --> 00:00:44,555 Shipping code in itself is zero outcome. 19 00:00:44,615 --> 00:00:47,865 Unless the system is being used. You emphasize a lot on 20 00:00:47,865 --> 00:00:50,345 accountability. How can you do that in such a 21 00:00:50,345 --> 00:00:52,445 way that still, nurtures psychological safety and people 22 00:00:52,445 --> 00:00:54,845 aren't afraid to actually admit things that they are doing. 23 00:00:54,875 --> 00:00:57,795 Every time we mess up, we diagnose the situation to its 24 00:00:57,795 --> 00:00:59,795 root cause. We do this for all processes. 25 00:00:59,855 --> 00:01:01,895 We try to figure out the accountable individual. 26 00:01:01,955 --> 00:01:04,614 Not to blame them and fire them, but really to learn to 27 00:01:04,614 --> 00:01:07,455 understand. We don't let seniority titles or 28 00:01:07,455 --> 00:01:11,359 org chart determine who's right. Right is like a factual 29 00:01:11,359 --> 00:01:14,015 statement, not one based on like social hierarchy. 30 00:01:14,075 --> 00:01:17,205 One of the most difficult things a startup faces is does a 31 00:01:17,205 --> 00:01:20,015 manager write code or not? As a manager, you can write 32 00:01:20,015 --> 00:01:21,189 code. I'm not gonna stop you if you 33 00:01:21,189 --> 00:01:23,793 write code. But I will only allow you to be 34 00:01:23,793 --> 00:01:26,115 measured in your performance by how you manage. 35 00:01:42,777 --> 00:01:44,289 Hello, guys. Welcome back to another new 36 00:01:44,289 --> 00:01:45,815 episode of the Tech Lead Journal podcast. 37 00:01:46,037 --> 00:01:49,367 Today, I have with me a CTO at Garner Health. 38 00:01:49,790 --> 00:01:52,835 His name is Michi Kono. Today, uh, we plan to talk 39 00:01:52,835 --> 00:01:55,575 about, a lot about, you know, how to grow and scale 40 00:01:55,575 --> 00:01:57,895 engineering team, building cultures, and things like that. 41 00:01:58,199 --> 00:02:01,444 Michi particularly also has some tips in terms of building some 42 00:02:01,444 --> 00:02:03,825 kind of a management system, engineering management, things 43 00:02:03,825 --> 00:02:06,197 like that. So, I'm really excited to have 44 00:02:06,197 --> 00:02:07,691 this conversation today with you, Michi. 45 00:02:07,691 --> 00:02:09,787 Welcome to the show. Great to be here. 46 00:02:09,787 --> 00:02:11,488 Thank you. Right. 47 00:02:11,488 --> 00:02:14,128 Michi, I always love to invite my guests first to maybe share a 48 00:02:14,128 --> 00:02:15,538 little bit more from your career. 49 00:02:15,568 --> 00:02:18,018 Any turning points that you think we can learn from you? 50 00:02:19,034 --> 00:02:21,903 Great. So early on in my career, I used 51 00:02:21,903 --> 00:02:25,311 to worked at a, a few startups and one of the things that I 52 00:02:25,311 --> 00:02:28,035 realized is that, you know, when I was looking for roles, I 53 00:02:28,035 --> 00:02:31,505 applied for a role once. And my friends had told me I'm 54 00:02:31,505 --> 00:02:34,576 not qualified for that role. And I ended up getting the role. 55 00:02:34,576 --> 00:02:36,806 It was for a lead engineer position, and it was actually a 56 00:02:36,806 --> 00:02:39,552 big turning point for me, because that startup did really 57 00:02:39,552 --> 00:02:41,866 well. I was one of the early hires and 58 00:02:41,866 --> 00:02:43,546 it sort of started my journey into management. 59 00:02:43,906 --> 00:02:46,880 And so I always tell people, don't reject yourself. 60 00:02:47,060 --> 00:02:50,793 Like that's someone else's job. What's the big harm in, you 61 00:02:50,793 --> 00:02:52,522 know, interviewing and not getting the role? 62 00:02:52,642 --> 00:02:55,462 And so I look back at that moment as like one of the most 63 00:02:55,462 --> 00:02:58,070 important decisions in my life. Cool. 64 00:02:58,160 --> 00:03:01,041 Yeah, that's very good thing to, you know, give as an advice, 65 00:03:01,041 --> 00:03:02,619 right? Because sometimes when we get 66 00:03:02,619 --> 00:03:05,432 rejected, we kind of like put yourself into the box, right? 67 00:03:05,432 --> 00:03:08,179 We think we are not good enough based on whatever feedback that 68 00:03:08,179 --> 00:03:09,878 people gave. So I think thanks for sharing 69 00:03:09,878 --> 00:03:11,755 that. So maybe if you can a little 70 00:03:11,755 --> 00:03:14,429 bit, dig a little bit deeper into that particular role, 71 00:03:14,429 --> 00:03:16,229 right? What makes you actually think 72 00:03:16,229 --> 00:03:19,060 that's a career turning points? Is it like the kind of growth, 73 00:03:19,060 --> 00:03:21,620 the kind of learning, the kind of opportunities and things like 74 00:03:21,620 --> 00:03:24,419 that? That role at that company wasn't 75 00:03:24,419 --> 00:03:28,541 a CTO role, but it was the first sort of IC engineering role. 76 00:03:28,961 --> 00:03:31,181 And so I ended up writing the whole code base. 77 00:03:31,181 --> 00:03:34,541 And because I was the first on the team, it sort of led to 78 00:03:34,541 --> 00:03:38,106 opportunities to lead the teams. And over time, eventually, I was 79 00:03:38,106 --> 00:03:40,947 promoted into management. And you know, that was very 80 00:03:40,947 --> 00:03:42,759 early in my career. You know, I'm not necessarily 81 00:03:42,759 --> 00:03:45,367 saying that was the right time for me at the time, but it was 82 00:03:45,367 --> 00:03:47,978 an opportunity that I wouldn't have had, and I certainly 83 00:03:47,978 --> 00:03:50,618 wouldn't have had that if I applied for the outside as a 84 00:03:50,618 --> 00:03:53,299 manager, for example. And so just taking a bet on 85 00:03:53,299 --> 00:03:54,739 myself there was pretty important. 86 00:03:55,467 --> 00:03:57,798 Yeah. So I think, uh, if I saw your 87 00:03:57,798 --> 00:04:00,286 career, right, so you started from this startup, it got 88 00:04:00,286 --> 00:04:02,926 acquired, and then you moved on into, you know, bigger companies 89 00:04:02,926 --> 00:04:05,433 like Capital One and then Meta and then Stripe. 90 00:04:05,673 --> 00:04:08,688 And now to Garner Health, right? So maybe from your journey, 91 00:04:08,688 --> 00:04:10,796 right, it seems like pretty unique and interesting, because 92 00:04:10,796 --> 00:04:13,683 you have seen, you know, being acquired in a startup, right? 93 00:04:13,683 --> 00:04:17,163 And working in a big tech, a big enterprise working in a, you 94 00:04:17,163 --> 00:04:19,656 know, I dunno like, financial services and now in health. 95 00:04:19,954 --> 00:04:23,051 Do you see some kind of, I dunno, like skill sets that you 96 00:04:23,051 --> 00:04:26,641 think you learn from when you are younger up to now that is 97 00:04:26,641 --> 00:04:29,590 still pretty relevant regardless of industries and companies? 98 00:04:30,416 --> 00:04:32,682 I think the most important theme, maybe we'll see this in 99 00:04:32,682 --> 00:04:35,966 other parts of my answers is around just communicating well. 100 00:04:36,682 --> 00:04:40,162 A lot of times as engineers, we can be really focused on like, 101 00:04:40,192 --> 00:04:43,106 how to write the code right? Or how to build the system, and 102 00:04:43,106 --> 00:04:45,774 then maybe not solve the problem when you finish building 103 00:04:45,774 --> 00:04:48,537 whatever you're building. And that was something that I 104 00:04:48,537 --> 00:04:51,672 learned early on, I think. First of all, I did some 105 00:04:51,672 --> 00:04:54,636 consulting at one point and that was very helpful 'cause that 106 00:04:54,636 --> 00:04:58,028 forces you to focus on like what the client wants. 107 00:04:58,589 --> 00:05:02,875 But also, uh, I remember, so yes, my startup was acquired at 108 00:05:02,875 --> 00:05:04,592 one point and I joined Capital One. 109 00:05:05,132 --> 00:05:07,670 And it was interesting because Capital One was kind of in the 110 00:05:07,670 --> 00:05:09,472 middle of acquiring a lot of startups at the time. 111 00:05:09,984 --> 00:05:12,110 Little acquisitions, not, not huge ones, but they were just 112 00:05:12,110 --> 00:05:13,774 kind of trying out the whole acquisition thing. 113 00:05:14,478 --> 00:05:18,031 And there was a small community of us, sort of people that were 114 00:05:18,031 --> 00:05:20,492 acquired. We were all about the same 115 00:05:20,492 --> 00:05:22,391 title. We all were about the same 116 00:05:22,391 --> 00:05:23,963 company size, same org level size. 117 00:05:24,323 --> 00:05:27,367 But I remember speaking to most of my peers and all of them were 118 00:05:27,367 --> 00:05:30,297 just waiting to quit. They were all just, they were 119 00:05:30,297 --> 00:05:33,813 just there to hang out and then leave in 18 months or whatever 120 00:05:33,813 --> 00:05:35,906 it was. And I remember thinking 121 00:05:35,906 --> 00:05:39,207 distinctly, and I told my team, hey, you know, we're here 122 00:05:39,207 --> 00:05:42,111 regardless for the next few years, so let's make the best of 123 00:05:42,111 --> 00:05:44,103 it. Let's commit to being here and 124 00:05:44,103 --> 00:05:46,613 understand what it means to be in a bigger company. 125 00:05:46,913 --> 00:05:48,533 It's not like being in a small company. 126 00:05:48,773 --> 00:05:51,413 There's skill sets here that we probably don't appreciate, and 127 00:05:51,413 --> 00:05:54,233 so let's build on that. And so that was, um, and I would 128 00:05:54,233 --> 00:05:57,089 say that's like maybe a second big learning was really 129 00:05:57,089 --> 00:06:01,763 committing to a situation and, you know, embracing the things 130 00:06:01,763 --> 00:06:04,159 that seem kind of bad from the outside and making sure that you 131 00:06:04,159 --> 00:06:06,356 understand what, you know, maybe if it is bad, you get to 132 00:06:06,356 --> 00:06:10,197 understand what makes it bad. Yeah, i've heard about, that 133 00:06:10,197 --> 00:06:10,820 huh? Sorry. 134 00:06:11,159 --> 00:06:13,588 Like, it wasn't bad. I was there for five years, so 135 00:06:13,588 --> 00:06:15,771 it was a great employer, just in case people might misinterpret 136 00:06:15,771 --> 00:06:19,046 what I was said there. Yeah, I've often heard about 137 00:06:19,046 --> 00:06:21,391 this story, right, when people got acquired. 138 00:06:21,391 --> 00:06:24,485 So there was, they would have to spend like a certain amount of 139 00:06:24,485 --> 00:06:27,685 years, let's say one year, two years, right, in order to serve 140 00:06:27,685 --> 00:06:29,905 within that company. And I always heard that people 141 00:06:29,905 --> 00:06:32,242 think, you know, it's like the free time for them, right? 142 00:06:32,332 --> 00:06:35,932 You just roam around, you know, do maybe less and, you know, 143 00:06:35,932 --> 00:06:39,058 think of your next adventure. So I think thanks for, again, 144 00:06:39,058 --> 00:06:41,442 sharing this important advice that probably you should commit, 145 00:06:41,442 --> 00:06:44,577 right, whatever journey that you are in the company, right, and 146 00:06:44,577 --> 00:06:48,855 try to make some contributions. So today the first topic that we 147 00:06:48,855 --> 00:06:52,022 want to talk about is about scaling and growing engineering 148 00:06:52,022 --> 00:06:54,400 teams. I think now that you are working 149 00:06:54,400 --> 00:06:57,167 in Garner Health, you are experiencing a rapid or 150 00:06:57,167 --> 00:06:58,892 hypergrowth in your company, right? 151 00:06:59,282 --> 00:07:02,207 So tell us first of all, like what kind of challenges that you 152 00:07:02,207 --> 00:07:04,742 typically see within such a company, right? 153 00:07:05,610 --> 00:07:08,596 I think that as you grow very quickly, people's roles evolve. 154 00:07:08,703 --> 00:07:12,051 That includes even my role. So like it's very different if 155 00:07:12,051 --> 00:07:15,495 one person is writing all the code, like I was at that one 156 00:07:15,495 --> 00:07:17,893 startup. And at some point, you'll cross 157 00:07:17,893 --> 00:07:19,987 a threshold where nobody does the whole code base. 158 00:07:19,987 --> 00:07:22,932 Nobody understands truly what's going on across the business. 159 00:07:23,057 --> 00:07:26,589 And so there are both, uh, knowledge, like sort of the way 160 00:07:26,589 --> 00:07:29,686 you operate changes that have to happen, but then there's also 161 00:07:29,686 --> 00:07:32,602 like a much higher overhead for like communication and 162 00:07:32,602 --> 00:07:36,291 coordination that happens. And if you don't adapt to that 163 00:07:36,291 --> 00:07:39,577 as it happens then, you know, your career could stagnate, or 164 00:07:39,577 --> 00:07:42,093 in some cases, you know, you won't make it in the company. 165 00:07:42,363 --> 00:07:45,903 And I've seen this pattern, you know, usually the ICs tend to 166 00:07:45,903 --> 00:07:48,412 survive that, because they can just become a master of the 167 00:07:48,412 --> 00:07:51,863 domain that they started with. But a lot of times, managers who 168 00:07:51,863 --> 00:07:55,019 are managing those teams can't sort of shed the skin of what 169 00:07:55,019 --> 00:07:57,515 they were good at in the previous role and adapt to the 170 00:07:57,515 --> 00:08:00,038 new circumstances that the company's in. 171 00:08:00,393 --> 00:08:02,133 And there's some rules of thumb for that. 172 00:08:02,133 --> 00:08:05,073 But the way I think about it is each new layer of management you 173 00:08:05,073 --> 00:08:09,219 add substantially changes the job required for success in that 174 00:08:09,219 --> 00:08:11,343 role. And so sometimes people will 175 00:08:11,343 --> 00:08:13,733 say, oh, great, I'm managing managers now and they're 176 00:08:13,733 --> 00:08:15,733 focusing on learning that skillset, but they haven't 177 00:08:15,733 --> 00:08:18,060 realized that the whole environment of how they operate 178 00:08:18,060 --> 00:08:19,623 with their peers, everything has changed. 179 00:08:20,013 --> 00:08:23,013 And so this happens very rapidly over the course of, you know, it 180 00:08:23,013 --> 00:08:25,076 could be 12 months, six months even sometimes, right, if you 181 00:08:25,076 --> 00:08:27,787 think of a doubling, a doubling out of a layer. 182 00:08:28,377 --> 00:08:30,869 And I've seen that like normally, you get years, right, 183 00:08:30,869 --> 00:08:33,384 in a corporate environment you can spend years working and then 184 00:08:33,384 --> 00:08:35,557 get promoted, and you can kind of see someone else doing the 185 00:08:35,557 --> 00:08:37,905 job and you kind of copy them. You don't get that. 186 00:08:38,166 --> 00:08:40,668 Like this happens almost in real time and everyone together is 187 00:08:40,668 --> 00:08:43,628 getting a new job. I always tell people, like my 188 00:08:43,628 --> 00:08:47,411 old manager, Nick, the CEO. His job substantially changes 189 00:08:47,411 --> 00:08:50,742 probably every year. And so the job that he was doing 190 00:08:50,742 --> 00:08:52,870 a year ago, all of his lieutenants have to do now. 191 00:08:53,510 --> 00:08:56,656 And so in some extent, we can copy each other. 192 00:08:56,986 --> 00:08:59,296 But sometimes the leader that you're copying from may not make 193 00:08:59,296 --> 00:09:01,901 it. And so that sort of mindset that 194 00:09:01,901 --> 00:09:04,876 you have to constantly reinvent yourself, I think is a cool 195 00:09:04,876 --> 00:09:06,166 opportunity for being in a startup. 196 00:09:06,436 --> 00:09:08,839 But it's also sort of the risk and reward of the situation. 197 00:09:09,634 --> 00:09:12,320 Yeah, this is, I think, one of the main struggles that I can 198 00:09:12,320 --> 00:09:15,334 see often, you know, people in this type of companies, right? 199 00:09:15,334 --> 00:09:18,444 I personally have experienced working in such a scaleup 200 00:09:18,444 --> 00:09:20,949 hypergrowth company, right? And it's always difficult to, 201 00:09:20,949 --> 00:09:24,036 you know, continuously evolve yourself in such a rapid time, 202 00:09:24,036 --> 00:09:25,961 right? So for example, maybe this year 203 00:09:25,961 --> 00:09:27,611 you're like, I dunno, 50 engineers. 204 00:09:27,791 --> 00:09:30,007 So maybe next year, it could be double and more than that, 205 00:09:30,007 --> 00:09:31,416 right? Especially as the business grow 206 00:09:31,416 --> 00:09:35,020 not so much in terms of just the size of people, but the size of 207 00:09:35,020 --> 00:09:36,582 problems, right? The domains that you're dealing 208 00:09:36,582 --> 00:09:38,762 with. The size of maybe, I dunno, like 209 00:09:38,762 --> 00:09:40,012 solutions, services and all that. 210 00:09:40,253 --> 00:09:42,412 And I think the size of counterparties as well. 211 00:09:42,472 --> 00:09:45,253 So there might be new leaders, business leaders, tech leaders 212 00:09:45,493 --> 00:09:46,873 that are also hired along the way. 213 00:09:47,443 --> 00:09:50,271 So I think if you can give maybe some tips for people who are 214 00:09:50,271 --> 00:09:54,073 experiencing this kind of stage, right, where they have to keep 215 00:09:54,073 --> 00:09:56,587 reinventing themselves, you know, over a certain period of 216 00:09:56,587 --> 00:09:59,143 time, what will be your best advice for them to actually 217 00:09:59,143 --> 00:10:03,394 learn and grow along the way? I think the first thing is to 218 00:10:03,394 --> 00:10:05,543 never be complacent of your past success. 219 00:10:05,939 --> 00:10:08,739 Like I said, the people who are most likely to not make it in 220 00:10:08,739 --> 00:10:10,619 these situations are actually the managers. 221 00:10:11,289 --> 00:10:13,293 Maybe I could be wrong. We can pull some numbers and 222 00:10:13,293 --> 00:10:15,936 look at it, but that's just my two cents on it, is that their 223 00:10:15,936 --> 00:10:18,943 job changes the most. And so you have to operate from 224 00:10:18,943 --> 00:10:21,243 the mindset, you're always basically sort of auditioning 225 00:10:21,243 --> 00:10:23,917 for your job, not because you're bad, but because the job that 226 00:10:23,917 --> 00:10:25,939 you're auditioning for is constantly changing and it's 227 00:10:25,939 --> 00:10:29,208 literally a moving bar. And so you just have to stay 228 00:10:29,208 --> 00:10:32,103 humble to what you're good at and say, hey, you know what I'm 229 00:10:32,103 --> 00:10:35,303 good at last month is no longer. And then don't forget that we're 230 00:10:35,303 --> 00:10:38,601 also in a technology org. So in parallel to that, the 231 00:10:38,601 --> 00:10:42,285 industry is also shifting. And a typical vast period for 232 00:10:42,285 --> 00:10:44,013 someone might be four years, right? 233 00:10:44,013 --> 00:10:46,293 So in four years, tech changes enormously. 234 00:10:46,343 --> 00:10:50,169 Like four years ago, I wanna say ChatGPT had just come out, you 235 00:10:50,169 --> 00:10:52,357 know, there was no such thing as AI, right? 236 00:10:52,357 --> 00:10:54,769 As a, or at least, it's at scale in the industry. 237 00:10:55,378 --> 00:10:58,294 Uh, and then four more years prior to that, like ML was 238 00:10:58,294 --> 00:11:00,493 hardly being utilized at scale in most companies, right? 239 00:11:00,493 --> 00:11:02,673 That was like if you were doing ML eight years ago, you were, 240 00:11:02,673 --> 00:11:05,930 you were a hot startup. And so I think in every company, 241 00:11:05,930 --> 00:11:07,736 you have to balance both those things. 242 00:11:07,736 --> 00:11:10,284 You need to be innovating. And so just in the same way that 243 00:11:10,284 --> 00:11:12,652 when you first joined the company early on, the tech 244 00:11:12,652 --> 00:11:15,665 engineering systems that you had overseen were state of the art 245 00:11:15,665 --> 00:11:19,322 in four years or not. And so either you're part of 246 00:11:19,322 --> 00:11:21,706 reinventing those systems or you're part of the problem. 247 00:11:22,256 --> 00:11:25,136 And so in those ways, like you have to just constantly be 248 00:11:25,136 --> 00:11:27,305 looking at yourself and the systems that you manage 249 00:11:27,305 --> 00:11:30,062 critically. Yeah, I can, uh, actually relate 250 00:11:30,062 --> 00:11:31,828 to that thing that you just mentioned, right? 251 00:11:31,828 --> 00:11:34,002 So in the past in that previous company, right? 252 00:11:34,002 --> 00:11:36,522 We do have a solution, right, that we built, you know, from 253 00:11:36,522 --> 00:11:39,342 the, from zero, let's say. And then over the time, right? 254 00:11:39,342 --> 00:11:42,166 It brings business, it keeps the company growing, but at certain 255 00:11:42,166 --> 00:11:45,266 point in time, it actually creates a lot of bottlenecks, 256 00:11:45,266 --> 00:11:46,906 right? Maybe challenges in terms of 257 00:11:46,906 --> 00:11:49,126 growth. Challenges if we wanna even keep 258 00:11:49,126 --> 00:11:52,761 on scaling it up, right? So I think as the manager or 259 00:11:52,761 --> 00:11:54,966 people who have experienced that before, right? 260 00:11:55,146 --> 00:11:57,846 I think it's really important to actually know, you know, how to 261 00:11:57,846 --> 00:12:00,726 evolve your system, how to evolve your team such that you 262 00:12:00,726 --> 00:12:04,254 can keep on continue scaling. And you don't slow down in terms 263 00:12:04,254 --> 00:12:07,326 of pace, which is, uh, what I want to ask you next, right? 264 00:12:07,326 --> 00:12:11,022 Because typically when people grow, we think that we will just 265 00:12:11,022 --> 00:12:15,066 indefinitely grow so fast. But I think one important 266 00:12:15,066 --> 00:12:18,690 phenomenon that always happen is that as you scale larger and 267 00:12:18,690 --> 00:12:21,552 larger, actually typically, the engineering delivery or maybe 268 00:12:21,552 --> 00:12:24,249 the output right, becomes much, much slower. 269 00:12:24,609 --> 00:12:27,213 So maybe from your point of view, do you experience the same 270 00:12:27,213 --> 00:12:29,370 thing? And if so, what typically are 271 00:12:29,370 --> 00:12:33,339 the root causes? So yes, delivery can slow, 272 00:12:33,339 --> 00:12:36,708 although I think the true slowness comes when you're much 273 00:12:36,708 --> 00:12:39,165 larger, where the average person just doesn't have an 274 00:12:39,165 --> 00:12:40,785 understanding of what the big picture is. 275 00:12:41,205 --> 00:12:44,883 But I think that it is still, again, sort of the inability to 276 00:12:44,883 --> 00:12:46,665 reinvent yourself plays a role for sure. 277 00:12:47,442 --> 00:12:50,202 Maybe triaging less the symptoms, but more the 278 00:12:50,202 --> 00:12:51,926 incentives I think are important. 279 00:12:51,926 --> 00:12:56,347 So like a lot of times when you're really small, everyone 280 00:12:56,347 --> 00:12:57,408 understands what they're supposed to do. 281 00:12:57,498 --> 00:12:58,998 You may not even have to assign work. 282 00:12:58,998 --> 00:13:00,618 People just sort of do the right thing, right? 283 00:13:01,068 --> 00:13:04,162 And then as you get larger, people no longer can do that or 284 00:13:04,162 --> 00:13:06,228 at least very few people can do that, right? 285 00:13:06,928 --> 00:13:10,008 And so I was, one of the things I saw at Meta in particular that 286 00:13:10,048 --> 00:13:15,006 I really liked, inspired maybe by, was that they ruthlessly 287 00:13:15,006 --> 00:13:16,941 hold teams accountable to outcomes. 288 00:13:17,719 --> 00:13:20,799 And so what I mean by that is, you know, I've worked at 289 00:13:20,799 --> 00:13:23,096 employers, most employers, in fact, where you've got this big 290 00:13:23,096 --> 00:13:25,417 deadline, you know, let's call it end of the year, you're 291 00:13:25,417 --> 00:13:28,493 working towards it. And you'll celebrate getting the 292 00:13:28,493 --> 00:13:31,168 coding to production. Say, hey, we shipped it, right, 293 00:13:31,168 --> 00:13:33,016 and that's your, that's your success finish line. 294 00:13:33,016 --> 00:13:35,926 And so maybe you ship it on December 31st and you win. 295 00:13:36,046 --> 00:13:38,416 Like and then you all get promotions and raises, right? 296 00:13:38,836 --> 00:13:42,796 But at Meta, that would be considered under expectations. 297 00:13:43,216 --> 00:13:45,676 So they look at the outcome. So what's the outcome that 298 00:13:45,676 --> 00:13:48,358 you're doing for the business? And so shipping code in itself 299 00:13:48,358 --> 00:13:49,456 is zero outcome. No. 300 00:13:49,726 --> 00:13:52,176 Unless the system is being used, right? 301 00:13:52,176 --> 00:13:54,586 So I remember a situation where we shipped something to 302 00:13:54,586 --> 00:13:56,680 production. There was no traffic on it, but 303 00:13:56,680 --> 00:13:59,572 it was celebrated, right? But at Meta, I remember a 304 00:13:59,572 --> 00:14:02,722 situation where a colleague of mine had spent six months 305 00:14:02,722 --> 00:14:05,843 working on a system. They shipped it at global scale 306 00:14:05,843 --> 00:14:09,682 and it had, you know, a couple hundred users a month using it, 307 00:14:09,682 --> 00:14:11,807 which is considered like basically non-existent. 308 00:14:11,927 --> 00:14:14,390 And so even though they had worked six months to ship it and 309 00:14:14,390 --> 00:14:17,882 build it and deploy it and get it used, they ended up getting a 310 00:14:17,882 --> 00:14:20,129 below expectations rating, because the outcome they were 311 00:14:20,129 --> 00:14:23,011 being held to was that this metric had moved or whatever 312 00:14:23,011 --> 00:14:26,171 thing was in that team. And obviously that sounds 313 00:14:26,171 --> 00:14:27,677 terrible. Like you could spend six months 314 00:14:27,677 --> 00:14:30,014 deliver everything you're supposed to do, and yet you get 315 00:14:30,014 --> 00:14:32,348 below expectations. But what that did, if you zoom 316 00:14:32,348 --> 00:14:34,772 out and think about what people do in, you know, as individuals 317 00:14:34,772 --> 00:14:38,261 is when you get assigned work to work on this thing for the next 318 00:14:38,261 --> 00:14:40,199 six months, you will audit that. Be like, well, hold on. 319 00:14:40,259 --> 00:14:42,509 Is this actually something that'll get used? 320 00:14:42,689 --> 00:14:45,641 Is there a way I can ship something in two months, maybe 321 00:14:45,641 --> 00:14:48,077 test the waters and if it's bad I got four more months to work 322 00:14:48,077 --> 00:14:50,339 on something else. Like it really creates the 323 00:14:50,339 --> 00:14:53,249 incentives down at the individual level to really 324 00:14:53,249 --> 00:14:55,319 understand the picture. At least, you need to understand 325 00:14:55,319 --> 00:14:58,781 the big picture for your part of the work in a way that I think 326 00:14:58,781 --> 00:15:00,719 creates global outcomes that are optimized. 327 00:15:00,839 --> 00:15:03,909 And so I always thought about that of like there's gotta be 328 00:15:03,909 --> 00:15:06,997 some, like, I think Meta, maybe it does it to an extreme that 329 00:15:06,997 --> 00:15:09,901 can be quite stressful. But I, but I always thought like 330 00:15:09,901 --> 00:15:12,212 that's, that mentality of holding engineers accountable 331 00:15:12,212 --> 00:15:15,429 for the outcomes ultimately helps them make better decisions 332 00:15:15,429 --> 00:15:18,225 and really empowers them. There'll be times where I saw 333 00:15:18,225 --> 00:15:20,331 engineers who would say no to their own leadership 'cause they 334 00:15:20,331 --> 00:15:22,657 would say, hey, you're gonna put me in a situation that I don't 335 00:15:22,657 --> 00:15:24,531 want to be in. This is not okay. 336 00:15:24,771 --> 00:15:26,241 This is a bad idea, we shouldn't do it. 337 00:15:26,421 --> 00:15:28,041 And that kind of stuff you just don't see anywhere else. 338 00:15:28,041 --> 00:15:30,687 But that's, it comes from the incentive structures of be 339 00:15:30,687 --> 00:15:32,271 holding them accountable if it goes poorly. 340 00:15:33,402 --> 00:15:36,563 Yeah, that's exactly what I wanted to ask the next, right, 341 00:15:36,563 --> 00:15:39,730 because in many organizations, typically also sometimes the 342 00:15:39,730 --> 00:15:43,175 order or maybe the instructions are given from the top, right? 343 00:15:43,175 --> 00:15:45,395 The top feels that this strategy might work. 344 00:15:45,695 --> 00:15:47,255 You know, you just go and deliver. 345 00:15:47,315 --> 00:15:49,505 You know, here are the set of high level requirements. 346 00:15:49,655 --> 00:15:52,499 The teams need to translate that into whatever shape, and then 347 00:15:52,499 --> 00:15:54,875 they just deliver it no matter what the timeframe. 348 00:15:55,115 --> 00:15:58,509 So I think with your explanation just now, right, I think this 349 00:15:58,509 --> 00:16:01,221 probably could not happen because you know, sometimes, uh, 350 00:16:01,221 --> 00:16:04,495 I mean as engineers or maybe the product managers within the 351 00:16:04,495 --> 00:16:07,125 team, they know that sometimes these ideas are not working. 352 00:16:07,570 --> 00:16:09,625 And because of that, right, actually people are reluctant to 353 00:16:09,625 --> 00:16:12,297 work on it. So tell us maybe in terms of 354 00:16:12,297 --> 00:16:14,845 changing the culture within the company, what should the top 355 00:16:14,845 --> 00:16:17,757 management do much better in terms, I dunno, maybe give more 356 00:16:17,757 --> 00:16:20,562 independence or maybe the team should push back a little bit 357 00:16:20,562 --> 00:16:22,692 more to the management. I think this all comes back to 358 00:16:22,692 --> 00:16:24,912 the culture as well. Maybe a little bit here as well. 359 00:16:25,919 --> 00:16:30,669 One of the things that we emphasize here at Garner is to 360 00:16:30,669 --> 00:16:33,724 design your orgs well. Like you start with what does 361 00:16:33,724 --> 00:16:36,579 the shape of the team look like? Not who do you have and what, 362 00:16:36,579 --> 00:16:39,312 where do you put them? And so that's an alluring 363 00:16:39,312 --> 00:16:41,519 concept in a startup. You, you're a small team. 364 00:16:41,519 --> 00:16:44,954 You've got four people, like you give them each one thing and you 365 00:16:44,954 --> 00:16:46,004 know, you have an org chart, right? 366 00:16:46,454 --> 00:16:49,694 But as you grow, and it's easy to sort of keep that model. 367 00:16:49,694 --> 00:16:51,559 And so you've got this person in charge of something and they 368 00:16:51,559 --> 00:16:54,407 stay in charge of it. But what we start doing in the 369 00:16:54,407 --> 00:16:57,221 company as we've scaled and said, okay, hold on, where do we 370 00:16:57,221 --> 00:16:59,873 want to be? What are the boxes then the org 371 00:16:59,873 --> 00:17:02,641 chart look like? And then you put people in as 372 00:17:02,641 --> 00:17:05,387 best fit. And that might, what this does 373 00:17:05,387 --> 00:17:07,037 is it starts from first principles. 374 00:17:07,037 --> 00:17:09,587 And so there could be a reorg that gets caused by that. 375 00:17:09,587 --> 00:17:12,700 But the idea is that you're evolving your business based on 376 00:17:12,700 --> 00:17:15,767 what you actually need, not so much who you have right now. 377 00:17:15,767 --> 00:17:18,227 'Cause you can always hire more people to fit the right 378 00:17:18,317 --> 00:17:21,805 skillsets that you're missing. And so how this relates to your 379 00:17:21,805 --> 00:17:25,242 question is that org charts sort of, they dictate both 380 00:17:25,242 --> 00:17:27,942 incentives. They dictate like what people 381 00:17:27,942 --> 00:17:30,322 prioritize. And so if you create your org 382 00:17:30,322 --> 00:17:33,724 chart in such a way where teams don't have autonomy, you create 383 00:17:33,724 --> 00:17:36,530 huge problems. I guess like an extreme example 384 00:17:36,530 --> 00:17:39,637 is each different type of engineer reports to a different 385 00:17:39,637 --> 00:17:41,916 manager, right? So you've got like a QA 386 00:17:41,916 --> 00:17:44,598 engineer, a frontend engineer, a backend engineer, and they all 387 00:17:44,598 --> 00:17:45,842 report to different managers, right? 388 00:17:46,691 --> 00:17:49,222 So you've created like a very complex matrix organization, and 389 00:17:49,222 --> 00:17:51,214 that makes sense. Like your marketers are 390 00:17:51,214 --> 00:17:53,678 marketing, your... sales people are in sales, obviously. 391 00:17:54,008 --> 00:17:56,426 But the problem is that if you're not careful with how you 392 00:17:56,426 --> 00:17:59,594 structure that, you're creating overhead in a way that makes 393 00:17:59,594 --> 00:18:01,666 coordination difficult. And more importantly, different 394 00:18:01,666 --> 00:18:03,398 managers might prioritize different things. 395 00:18:03,818 --> 00:18:06,612 And so you can imagine the extreme world where two managers 396 00:18:06,612 --> 00:18:09,742 are just incentivized to do two different things, and then you 397 00:18:09,742 --> 00:18:13,146 have bad outcomes occur. Yeah, so I think I've heard this 398 00:18:13,146 --> 00:18:15,411 from James Stanier as well regarding org chart, right? 399 00:18:15,411 --> 00:18:19,633 I think that's the one of the most powerful tool in fact to 400 00:18:19,633 --> 00:18:21,381 actually shape the culture, that's the first thing. 401 00:18:21,501 --> 00:18:23,901 And second thing is about, you know, the communication, the 402 00:18:23,901 --> 00:18:26,394 flow within, you know, one team versus the other. 403 00:18:26,694 --> 00:18:29,463 And matrix organization, uh, even though some companies are 404 00:18:29,463 --> 00:18:32,108 still adopting it, probably that's also kind of like 405 00:18:32,108 --> 00:18:34,736 creating a lot of tensions, because like different managers 406 00:18:34,736 --> 00:18:37,385 will be aligned to different incentives, different outcomes, 407 00:18:37,385 --> 00:18:39,196 right? And sometimes for the IC 408 00:18:39,196 --> 00:18:41,330 themselves having to report to two different managers, 409 00:18:41,330 --> 00:18:44,462 sometimes it's not easy and tricky to actually align what 410 00:18:44,462 --> 00:18:45,752 should be the work that they're doing. 411 00:18:46,112 --> 00:18:49,179 So I think org chart definitely is a much, much, play much 412 00:18:49,179 --> 00:18:51,114 bigger important role in any organization. 413 00:18:51,324 --> 00:18:54,600 So do you have like, I dunno, an evolution from a startup to a 414 00:18:54,600 --> 00:18:57,873 bigger company, a typical org chart that you would want to 415 00:18:57,873 --> 00:19:02,160 suggest people? Ideally, you want the manager to 416 00:19:02,160 --> 00:19:05,840 have as much autonomy as possible to deliver what they're 417 00:19:05,840 --> 00:19:08,372 working on. In other words, let's use that 418 00:19:08,372 --> 00:19:10,692 example of the frontend, backend, and QA person. 419 00:19:10,752 --> 00:19:13,969 Like you would just have a team that's responsible for shipping 420 00:19:13,969 --> 00:19:15,428 a feature. Some feature that you're working 421 00:19:15,428 --> 00:19:16,953 on. Let's say, it's update a button 422 00:19:16,953 --> 00:19:19,843 or whatever. And so they've, that manager 423 00:19:19,843 --> 00:19:22,773 owns the responsibility for designing the backend, designing 424 00:19:22,773 --> 00:19:25,908 the frontend and QA. Now, at some point that gets 425 00:19:25,908 --> 00:19:28,868 difficult cause the more further you go down, the more it's like 426 00:19:28,868 --> 00:19:30,213 different from like the button push. 427 00:19:30,903 --> 00:19:34,531 The idea is that you wanna have the business sort of decision 428 00:19:34,531 --> 00:19:38,466 making to be in one place. And so eventually you'll have 429 00:19:38,466 --> 00:19:40,985 some collaboration, but you wanna make those as minimal as 430 00:19:40,985 --> 00:19:44,721 possible in the places that are sort of not gonna have opposing 431 00:19:44,721 --> 00:19:46,990 interests, right? So if you have two different 432 00:19:46,990 --> 00:19:49,712 buttons being worked on, and you have one engineer, that frontend 433 00:19:49,712 --> 00:19:51,882 engineer managed by someone else, you've now created a 434 00:19:51,882 --> 00:19:53,894 situation where you're fighting over that one engineer, right? 435 00:19:53,894 --> 00:19:55,798 So you have to say, you know what, we're not gonna, we're 436 00:19:55,798 --> 00:19:58,600 gonna put that engineer in the team for button A and button B 437 00:19:58,600 --> 00:19:59,984 doesn't get a team. Or it's not. 438 00:20:00,012 --> 00:20:02,326 Or it's under the same manager or what, and that would cause a 439 00:20:02,326 --> 00:20:04,980 reorg, right? And so the idea is that you're 440 00:20:04,980 --> 00:20:07,689 designing from, first of all, like autonomy, but it's also 441 00:20:07,689 --> 00:20:09,924 like what the business needs and that could change your gear. 442 00:20:10,104 --> 00:20:12,366 And so you might see when you're like, oh, this one thing that 443 00:20:12,366 --> 00:20:15,274 used to be pretty clear and straightforward owned by one 444 00:20:15,274 --> 00:20:17,893 team, over the years, it's evolved and now two teams own 445 00:20:17,893 --> 00:20:19,566 it. And you might think like, well, 446 00:20:19,566 --> 00:20:21,294 maybe your options, you have two options. 447 00:20:21,294 --> 00:20:24,185 Either you acknowledge that it's shared and you create a 448 00:20:24,185 --> 00:20:27,304 platform. Or you would do a reorg and move 449 00:20:27,304 --> 00:20:29,634 that scope into that. It doesn't have to be reorg, I 450 00:20:29,634 --> 00:20:31,358 guess. We can move the project into one 451 00:20:31,358 --> 00:20:34,472 team and so that there's no longer a multi-team dependency. 452 00:20:34,772 --> 00:20:36,272 And that's greatly oversimplifying it. 453 00:20:36,272 --> 00:20:38,732 Like this is hard, like hard emotionally for people obviously 454 00:20:38,732 --> 00:20:40,322 cause you're changing people's work. 455 00:20:40,322 --> 00:20:42,512 You might like take someone's project away, that's their baby. 456 00:20:42,953 --> 00:20:46,051 But these decisions, if you don't do it, can really cause 457 00:20:46,051 --> 00:20:49,493 harm across the org, because everyone is sort of, you know, 458 00:20:49,493 --> 00:20:53,140 handcuffed to inertia. Yeah, I like the way that you 459 00:20:53,140 --> 00:20:55,285 emphasize, uh, you know, aligning in terms of autonomy 460 00:20:55,285 --> 00:20:56,815 and also like decision making, right? 461 00:20:56,815 --> 00:20:59,547 So try to make as much as possible decision making 462 00:20:59,547 --> 00:21:02,186 centralized in one place, right, instead of, you know, having two 463 00:21:02,186 --> 00:21:04,127 different managers, two different teams trying to align 464 00:21:04,127 --> 00:21:06,719 a certain priority. So I think in terms of culture 465 00:21:06,719 --> 00:21:09,507 as well, these kind of things, definitely not so easy to 466 00:21:09,507 --> 00:21:12,780 evolve, uh, especially when you hire so many people from 467 00:21:12,780 --> 00:21:15,775 outside, right, coming in. So, you know, growing 468 00:21:15,775 --> 00:21:19,213 engineering culture is always like a very hard thing to do, 469 00:21:19,213 --> 00:21:20,833 but it is a very important thing to do. 470 00:21:21,140 --> 00:21:24,006 So maybe in your CTO role currently at Garner Health, 471 00:21:24,006 --> 00:21:27,243 what's your tips and strategy to actually, you know, nurture a 472 00:21:27,243 --> 00:21:29,199 good engineering culture? And what kind of engineering 473 00:21:29,199 --> 00:21:30,264 culture that you're building anyway? 474 00:21:31,089 --> 00:21:33,429 So yeah, every company has a slightly different culture. 475 00:21:33,846 --> 00:21:36,816 For us at Garner, you know, we're very mission driven. 476 00:21:37,094 --> 00:21:40,964 Our website has these values, and, um, uh, I would say like 477 00:21:41,112 --> 00:21:42,938 they're not just marketing. Like we actually hold ourselves 478 00:21:42,938 --> 00:21:45,402 to them internally in a way that's very hard to explain to 479 00:21:45,402 --> 00:21:47,652 an external person. But we live these values. 480 00:21:47,712 --> 00:21:50,388 And so there's no room for politics. 481 00:21:50,583 --> 00:21:53,383 Politics is another way of saying human, human social 482 00:21:53,383 --> 00:21:55,123 interactions. Like it happens, but to the 483 00:21:55,123 --> 00:21:57,786 level of, you know, deception or anything like that, you know, or 484 00:21:57,786 --> 00:21:59,434 ego getting in the way of good outcomes. 485 00:21:59,824 --> 00:22:02,666 And so one of the things we do, for example, is every time we 486 00:22:02,666 --> 00:22:06,364 mess up, we diagnose the situation to its root cause. 487 00:22:06,815 --> 00:22:08,973 In engineering terms, you can think of that as like an 488 00:22:08,973 --> 00:22:11,518 incident review, but we do this for all processes. 489 00:22:12,268 --> 00:22:14,408 And the other thing is that it's not blameless. 490 00:22:14,440 --> 00:22:17,795 It's actually, like fully we try to figure out the accountable 491 00:22:17,795 --> 00:22:20,214 individual. It is not, not to blame them and 492 00:22:20,214 --> 00:22:22,113 fire them, but really to learn to understand. 493 00:22:22,113 --> 00:22:24,345 It's just like an incident. Like we need to understand why 494 00:22:24,345 --> 00:22:26,613 did team push the button that caused the outage? 495 00:22:26,613 --> 00:22:29,654 Not just that the button got pushed, because there's a reason 496 00:22:29,654 --> 00:22:32,537 behind that button push. Maybe the manager told them to 497 00:22:32,537 --> 00:22:34,685 do it or whatever. We need to understand that and 498 00:22:34,685 --> 00:22:37,517 so we can learn from it. That's probably the most unique 499 00:22:37,517 --> 00:22:40,140 part of our culture, in that it's, there's no room to hide 500 00:22:40,140 --> 00:22:42,396 about from this kind of stuff. We really make sure that 501 00:22:42,396 --> 00:22:45,079 everyone is operating in the clear and that we all understand 502 00:22:45,079 --> 00:22:48,935 what's going behind the things. And we don't let seniority 503 00:22:48,935 --> 00:22:52,345 titles or org chart determine who's right. 504 00:22:52,645 --> 00:22:57,346 Like right is like a factual statement, not one based on like 505 00:22:57,346 --> 00:23:00,566 social hierarchy. And I think that's probably the 506 00:23:00,566 --> 00:23:02,757 most important cultural thing that we do. 507 00:23:03,097 --> 00:23:06,712 And then I think also from, uh, this gets maybe the management 508 00:23:06,712 --> 00:23:10,850 stuff, but I think a good engineering culture is one that 509 00:23:10,850 --> 00:23:13,481 focuses on engineering. In other words, the ICs. 510 00:23:14,005 --> 00:23:16,600 You know, we, as managers, we oversee the work. 511 00:23:17,020 --> 00:23:19,450 We might guide folks and coach people. 512 00:23:19,750 --> 00:23:22,603 But ultimately the person on the ground writing the code is these 513 00:23:22,603 --> 00:23:24,786 engineers. And if you forget that fact, um, 514 00:23:24,786 --> 00:23:27,676 and you manage top down too much, then really bad outcomes 515 00:23:27,676 --> 00:23:30,353 will happen. I remember being at the bank and 516 00:23:30,353 --> 00:23:33,863 again, I love working at Capital One, but being at the bank, and 517 00:23:33,863 --> 00:23:36,195 there were times. There were explicit times I can 518 00:23:36,195 --> 00:23:39,873 remember where someone with a VP title would come into a meeting 519 00:23:39,873 --> 00:23:43,713 and say, hey, the API should be built this way, obviously. 520 00:23:43,773 --> 00:23:45,586 Everybody, you're all wrong. This is how we're gonna build 521 00:23:45,586 --> 00:23:47,151 it. You know, there was the, you 522 00:23:47,151 --> 00:23:49,333 know, you showed the executive leadership like two very good 523 00:23:49,333 --> 00:23:51,753 options, A or B, of like how to build something. 524 00:23:51,753 --> 00:23:53,763 And you always got option C. That was the joke. 525 00:23:53,913 --> 00:23:56,802 And option C was always suboptimal from an engineering 526 00:23:56,802 --> 00:23:59,508 perspective. And, you know, it was most of 527 00:23:59,508 --> 00:24:01,503 the time fine, like we'd figure out a way around it. 528 00:24:01,503 --> 00:24:04,341 But there was a couple times where I did literally tell that 529 00:24:04,341 --> 00:24:07,389 person, hey, you know, you said something that's like actually 530 00:24:07,389 --> 00:24:11,168 nonsensical and it's not right. But if I hadn't said that, it 531 00:24:11,168 --> 00:24:13,182 would just have autopilot been approved and everyone would've 532 00:24:13,182 --> 00:24:15,923 just followed orders and we wouldn't end up with the worst 533 00:24:15,923 --> 00:24:17,622 system. And so that's the extreme. 534 00:24:17,642 --> 00:24:20,740 We don't want that. And so you want a world where 535 00:24:20,740 --> 00:24:23,432 your best engineers are engineering the right systems. 536 00:24:23,612 --> 00:24:25,994 And I think the only counterbalance is really the 537 00:24:25,994 --> 00:24:27,587 business outcomes because sometimes building the best 538 00:24:27,587 --> 00:24:28,862 system might take you four years. 539 00:24:28,922 --> 00:24:31,262 You don't have four years. And so you have to balance that. 540 00:24:31,262 --> 00:24:34,406 So at that part, I understand, but don't let the manager's 20 541 00:24:34,406 --> 00:24:37,244 years of engineering experience overrule the two years of 542 00:24:37,244 --> 00:24:39,946 engineering experience, because that 20 years of experience was 543 00:24:39,946 --> 00:24:42,872 a decade ago, you know, and so it's out of date. 544 00:24:42,872 --> 00:24:45,102 And so the two year of experience person might actually 545 00:24:45,102 --> 00:24:47,868 be more correct, right? And so you have, we have to 546 00:24:47,868 --> 00:24:50,252 evaluate the ideas, the engineering ideas for what they 547 00:24:50,252 --> 00:24:53,102 are, not like the title of the person who said them. 548 00:24:53,634 --> 00:24:56,504 And sometimes even, as a CTO, I say stuff and I always tell 549 00:24:56,504 --> 00:24:58,414 people like, I could be totally wrong. 550 00:24:58,444 --> 00:25:01,793 I know I have this nice title, but don't take what I say as an 551 00:25:01,793 --> 00:25:03,966 order. It's just an idea. 552 00:25:04,176 --> 00:25:05,916 You are welcome to tell me I'm wrong. 553 00:25:05,916 --> 00:25:07,920 And I, and again, going back to right diagnosing root causes, 554 00:25:07,920 --> 00:25:10,646 the last thing I want is the diagnosis of the root cause to 555 00:25:10,646 --> 00:25:14,166 be because Michi said so. Yeah, thanks for sharing such a 556 00:25:14,166 --> 00:25:16,584 good story, right? So because in many typical 557 00:25:16,584 --> 00:25:18,588 organizations, sometimes, it's the role that actually 558 00:25:18,588 --> 00:25:21,546 determines, you know, the kind of work that you do, the kind of 559 00:25:21,546 --> 00:25:24,260 things that you prioritize. So I think the most important 560 00:25:24,260 --> 00:25:27,992 thing just now, when I heard you mention is also to emphasize a 561 00:25:27,992 --> 00:25:30,232 lot on engineers, right? So engineering culture means 562 00:25:30,232 --> 00:25:32,683 like, yeah, you really care about the engineers and you 563 00:25:32,683 --> 00:25:35,896 don't just, you know, manage them as a, as if they are just 564 00:25:35,896 --> 00:25:37,316 resources for you to do the work. 565 00:25:37,768 --> 00:25:40,870 I'm a little bit intrigued when you mentioned about, you know, 566 00:25:40,870 --> 00:25:44,114 like your kind of like root cause analysis is slightly more 567 00:25:44,114 --> 00:25:45,614 nuanced than the blameless, right? 568 00:25:45,614 --> 00:25:47,898 Because typically people talk about blameless, postmortem, 569 00:25:47,898 --> 00:25:49,810 root cause analysis and things like that. 570 00:25:49,840 --> 00:25:51,970 And you emphasize a lot on accountability. 571 00:25:52,300 --> 00:25:55,940 How can you do that, uh, in such a way that still, you know, 572 00:25:55,940 --> 00:25:58,900 nurtures psychological safety and people are not kind of like 573 00:25:58,900 --> 00:26:02,068 blamed or people aren't afraid to actually admit things that 574 00:26:02,068 --> 00:26:04,515 they're doing. Or maybe even like they are 575 00:26:04,515 --> 00:26:07,224 afraid to take actions simply, because they don't wanna be 576 00:26:07,224 --> 00:26:09,554 accountable at the end. Yeah. 577 00:26:09,945 --> 00:26:11,475 I'll start to say it's very difficult. 578 00:26:11,595 --> 00:26:15,622 You know, once you start using, if you had in incident reviews 579 00:26:15,622 --> 00:26:20,018 that always sort of said who did the thing, it can certainly 580 00:26:20,018 --> 00:26:22,641 cause problems over time, incentives where like, 581 00:26:22,641 --> 00:26:24,950 especially junior people, they're gonna be more less 582 00:26:24,950 --> 00:26:28,869 likely to speak up. I think a big part of that 583 00:26:28,869 --> 00:26:33,389 though is don't let this become a delegation of bad outcomes 584 00:26:33,389 --> 00:26:36,285 exercise, right? Like there'll be plenty of times 585 00:26:36,285 --> 00:26:39,380 where the root cause is literally because Michi said so. 586 00:26:39,380 --> 00:26:42,534 And so the learning is Michi needs to be more careful about 587 00:26:42,534 --> 00:26:45,810 what he says. Like I have done these diagnosis 588 00:26:45,810 --> 00:26:49,105 myself and taken accountability where I messed up. 589 00:26:49,555 --> 00:26:53,638 And so it's truly not meant to be, you know, it's all everyone 590 00:26:53,638 --> 00:26:56,661 else's fault exercise. And I think, as a leader, I have 591 00:26:56,661 --> 00:26:58,785 to step in and say, well, here's where how I contributed. 592 00:26:58,785 --> 00:27:02,882 Because there'll be very few cases, especially outside of 593 00:27:02,882 --> 00:27:05,732 incidents, there'll be very few cases where I'm fully blameless. 594 00:27:06,232 --> 00:27:08,704 There is some level of accountability I have, to like, 595 00:27:08,704 --> 00:27:11,294 it could be as simple as like, I'm the one that assigned the 596 00:27:11,294 --> 00:27:13,232 person and they were three weeks at the job, right? 597 00:27:13,232 --> 00:27:15,934 Like it could be me that assigned it or it could be me 598 00:27:15,934 --> 00:27:18,589 that set the conditions for success, but it was the wrong 599 00:27:18,589 --> 00:27:20,697 conditions. So I think by demonstrating that 600 00:27:20,697 --> 00:27:24,645 culture, like I said, we live it in this company at the top of 601 00:27:24,645 --> 00:27:26,183 the organization, including with Nick. 602 00:27:26,793 --> 00:27:29,889 I think that it sets the tone that this is okay, that you 603 00:27:29,889 --> 00:27:32,073 don't, you're not gonna get fired just 'cause you mess up. 604 00:27:32,325 --> 00:27:34,344 As long as the conversation that's happening is how we're 605 00:27:34,344 --> 00:27:36,195 gonna learn, then everyone will be fine. 606 00:27:36,616 --> 00:27:40,096 Whereas like hiding the truth would definitely be bad. 607 00:27:40,529 --> 00:27:43,259 Or using politics and ego and not admitting that would be bad. 608 00:27:43,259 --> 00:27:46,839 But you could literally say, I did it because in certain 609 00:27:46,839 --> 00:27:49,929 political reason, you could say that and say, my learning is to 610 00:27:49,929 --> 00:27:53,374 be less political, right? So I think that it's all about 611 00:27:53,374 --> 00:27:56,234 demonstrating those values for everybody and making sure to 612 00:27:56,234 --> 00:27:58,524 show grace when it happens and focusing more on the learnings 613 00:27:58,524 --> 00:28:02,615 and less about punishing people. Yeah, so I think, again, the 614 00:28:02,615 --> 00:28:05,333 leadership sets the tone, right? So if the leaders also kinda 615 00:28:05,333 --> 00:28:08,148 like admit mistakes, be part of the, you know, the root cause 616 00:28:08,148 --> 00:28:10,713 analysis and try to learn for themselves as well and train 617 00:28:10,713 --> 00:28:13,463 people, I think that's really important thing to build that 618 00:28:13,463 --> 00:28:16,003 culture. Another thing is like whenever 619 00:28:16,003 --> 00:28:19,935 people make mistakes, how do you actually make them learn from 620 00:28:19,935 --> 00:28:22,687 the mistakes? Do you put it as part of their 621 00:28:22,687 --> 00:28:25,675 performance review? Do you put that as part of the 622 00:28:25,675 --> 00:28:28,040 team performance thing? How do you actually ensure 623 00:28:28,040 --> 00:28:29,708 people learn and grow from the mistakes? 624 00:28:30,859 --> 00:28:33,832 I don't know if I'd couple mistakes with performance that 625 00:28:33,832 --> 00:28:35,932 tightly. Like we all make mistakes. 626 00:28:36,170 --> 00:28:39,371 I make mistakes every week for whatever the, whatever the topic 627 00:28:39,371 --> 00:28:41,049 is. Even if you really experienced, 628 00:28:41,049 --> 00:28:42,704 you'll still make mistakes. You know, otherwise, there'd be 629 00:28:42,704 --> 00:28:44,505 no car crashes. We spend lots of time driving 630 00:28:44,505 --> 00:28:45,883 cars, yet, we still get accidents, right? 631 00:28:45,903 --> 00:28:48,363 So that's inevitable. There's gonna be mistakes. 632 00:28:48,423 --> 00:28:50,043 And maybe, so let's start with that. 633 00:28:50,043 --> 00:28:52,893 Like performance reviews are not a place to call out mistakes. 634 00:28:53,569 --> 00:28:57,089 Now if it's a thematic problem, right, versus instance. 635 00:28:57,089 --> 00:28:59,388 In other words, you make the same mistake over and over and 636 00:28:59,388 --> 00:29:01,354 over. You're always, you know, hitting 637 00:29:01,354 --> 00:29:04,225 cars when you're reversing. Then we should talk about taking 638 00:29:04,225 --> 00:29:06,607 away your license, right? So only then would I consider it 639 00:29:06,607 --> 00:29:08,683 a performance problem. Like I'd be very clear, hey, 640 00:29:08,683 --> 00:29:10,279 this is actually like the eighth time. 641 00:29:10,279 --> 00:29:13,002 This is a performance problem versus, I mean, we all make 642 00:29:13,002 --> 00:29:15,237 mistakes. And performance should also be 643 00:29:15,237 --> 00:29:17,635 about success. Like there's just as much. 644 00:29:17,815 --> 00:29:20,052 In fact, maybe even more so performance reviews, you should 645 00:29:20,052 --> 00:29:21,835 think about the outcomes that were good. 646 00:29:22,563 --> 00:29:25,123 You shouldn't even get to the performance review process step 647 00:29:25,123 --> 00:29:28,653 if the person is bad, because they should have been coached. 648 00:29:28,713 --> 00:29:30,689 And if they're not learning, like you probably don't wanna 649 00:29:30,689 --> 00:29:32,511 get to that point where you're having a performance 650 00:29:32,511 --> 00:29:34,623 conversation in the performance review, if that makes sense. 651 00:29:34,623 --> 00:29:36,677 Like you want to start, like you want to make sure people that 652 00:29:36,677 --> 00:29:39,213 are not performing are not there much sooner. 653 00:29:39,793 --> 00:29:42,043 Cause you only do performance reviews once or twice a year. 654 00:29:42,523 --> 00:29:45,043 But so if you are getting to that point with someone, then 655 00:29:45,043 --> 00:29:46,213 they've probably been successful. 656 00:29:46,963 --> 00:29:49,273 And I think that the conversation needs to be, what 657 00:29:49,273 --> 00:29:50,933 are the things that they've done well? 658 00:29:50,933 --> 00:29:52,353 First of all, recognizing that work. 659 00:29:52,883 --> 00:29:54,902 And then you can have the coaching conversation of, you 660 00:29:54,902 --> 00:29:57,623 know, here's the gap that here's the learning so we, we can do 661 00:29:57,623 --> 00:29:59,098 better. And then here's where you're 662 00:29:59,098 --> 00:30:01,117 like, understand the career goals and say, hey, where, 663 00:30:01,117 --> 00:30:03,353 what's the gap between where you are today and your career goals. 664 00:30:04,280 --> 00:30:07,364 So I think it's very important, like you should not wait, uh, 665 00:30:07,364 --> 00:30:10,415 when you see some people, you know, kind of like maybe, I 666 00:30:10,415 --> 00:30:12,594 dunno, misaligned or make mistakes or repetitively make 667 00:30:12,594 --> 00:30:14,595 mistakes, right? You don't wait until the 668 00:30:14,595 --> 00:30:15,960 performance review cycle happening, right? 669 00:30:15,960 --> 00:30:19,260 So you actually coach them, maybe put in one-on-one agenda, 670 00:30:19,260 --> 00:30:22,372 like try to actually understand the root cause and try to 671 00:30:22,372 --> 00:30:25,380 actually help them to perform better in their role, right? 672 00:30:25,380 --> 00:30:26,730 So I think thanks for emphasizing that. 673 00:30:27,170 --> 00:30:30,060 One other things that I always get intrigued when people say 674 00:30:30,060 --> 00:30:32,552 mission driven, right? Because almost every company, 675 00:30:32,552 --> 00:30:35,400 mostly startups, right, they think they are mission driven. 676 00:30:35,520 --> 00:30:38,340 They are, they live by the values that they are, you know, 677 00:30:38,340 --> 00:30:40,764 like they put in their websites, products and all that. 678 00:30:41,124 --> 00:30:43,494 How do you actually become really mission driven? 679 00:30:43,554 --> 00:30:46,434 Is there something in particular that we can learn from how you 680 00:30:46,434 --> 00:30:49,229 run Garner Health? I was just thinking to myself, 681 00:30:49,229 --> 00:30:51,391 is the average person mission driven in a startup. 682 00:30:51,841 --> 00:30:54,601 I feel like there's a good chunk of people that are just excited 683 00:30:54,601 --> 00:30:58,643 by the adrenaline and the money. But we, at Garner, obviously, 684 00:30:58,643 --> 00:31:03,132 like the average employee here is more mission driven than ever 685 00:31:03,132 --> 00:31:05,256 encountered at any other company, most. 686 00:31:05,461 --> 00:31:08,401 It was not uncommon for me to speak to an engineer in the past 687 00:31:08,401 --> 00:31:10,708 and they would say in the interview process at the end of 688 00:31:10,708 --> 00:31:12,042 it. So what do you guys do here? 689 00:31:12,680 --> 00:31:15,121 I saw that you guys do Python. I love Python, so I know I'm 690 00:31:15,121 --> 00:31:17,104 here, but what do you, what do you guys do with that? 691 00:31:17,864 --> 00:31:20,665 That was very common for someone to not understand the business 692 00:31:20,665 --> 00:31:23,071 model of a company, especially if it's a complex business model 693 00:31:23,071 --> 00:31:26,189 like Garner or healthcare tends to be complex in terms of the 694 00:31:26,189 --> 00:31:28,932 financial model. But it, but here, it's been 695 00:31:28,932 --> 00:31:30,532 incredible to see how mission-driven everybody, 696 00:31:30,532 --> 00:31:33,198 everyone really cares cause healthcare is very personal. 697 00:31:33,715 --> 00:31:36,410 I would argue it's, in that sense, it's quite easy because 698 00:31:36,410 --> 00:31:40,719 what we do is, you know, our company's technology, if it's 699 00:31:40,719 --> 00:31:43,119 successful, ultimately helps people get better healthcare. 700 00:31:43,119 --> 00:31:45,264 So it's a very easy mission to sort of get people to aligned 701 00:31:45,264 --> 00:31:47,916 with. We're not trying to like, make 702 00:31:47,916 --> 00:31:51,078 insurance premiums, you know, 20% cheaper or like improve 703 00:31:51,078 --> 00:31:52,903 margins at a hospital. Like those things are a little 704 00:31:52,903 --> 00:31:54,833 bit harder to relate to. I think personally, even though, 705 00:31:54,833 --> 00:31:56,773 yeah, you're making healthcare better, this is literally about 706 00:31:56,773 --> 00:31:59,530 giving you a better doctor. So it's, I found that it's been 707 00:31:59,530 --> 00:32:01,738 an easy sell to sell people on the mission. 708 00:32:02,318 --> 00:32:04,298 And make sure that, again, going back to root causes, really 709 00:32:04,298 --> 00:32:06,674 making sure that we're focused on the mission and not so much 710 00:32:06,674 --> 00:32:08,798 on like the ego and the politics of a situation. 711 00:32:09,038 --> 00:32:10,328 I hope that answers your question. 712 00:32:11,184 --> 00:32:14,664 So one as aspect as, uh, maybe an extension from what you just 713 00:32:14,664 --> 00:32:16,548 mentioned, right? How do you actually ensure that 714 00:32:16,548 --> 00:32:19,952 the day to day, you know, the employees that work within the 715 00:32:19,952 --> 00:32:23,417 company always get aligned to that mission, or maybe that 716 00:32:23,447 --> 00:32:25,157 values that you have in the companies. 717 00:32:25,471 --> 00:32:27,661 Do you actually put it as part of performance review? 718 00:32:27,661 --> 00:32:30,517 Do you actually put it as part of, I dunno, like your 719 00:32:30,517 --> 00:32:32,561 communication often that, hey, we have these values, you have 720 00:32:32,561 --> 00:32:35,189 to embody it, and this is how you show examples and things 721 00:32:35,189 --> 00:32:37,136 like that. Maybe some practical tips, how 722 00:32:37,136 --> 00:32:40,020 do you actually align your mission and vision and the 723 00:32:40,020 --> 00:32:41,567 values within the day-to-day work? 724 00:32:42,545 --> 00:32:45,560 So we do. I'm reluctant to say this 725 00:32:45,560 --> 00:32:47,725 because the performance management process evolves as 726 00:32:47,725 --> 00:32:50,757 quickly as the company. So, you know, I've been here for 727 00:32:50,757 --> 00:32:53,345 a little less than a year, and my understanding is it changes 728 00:32:53,345 --> 00:32:55,728 every year. That's how it was at other 729 00:32:55,728 --> 00:32:57,095 companies. That's how it was even at 730 00:32:57,095 --> 00:33:00,479 Stripe, it changed every year. So I'll reluctantly say yes. 731 00:33:00,532 --> 00:33:04,126 Our performance management process does incorporate the 732 00:33:04,126 --> 00:33:08,356 values and literally ask, does the person operate under those 733 00:33:08,356 --> 00:33:10,716 value assumptions? And there's a rubric and we 734 00:33:10,716 --> 00:33:12,308 author it and change it year to year. 735 00:33:12,948 --> 00:33:17,257 But there's a, there's core tenets, uh, on the website and 736 00:33:17,257 --> 00:33:18,938 internally of like what the values are. 737 00:33:18,938 --> 00:33:21,878 And then we think about like for a given level of seniority, what 738 00:33:21,878 --> 00:33:25,906 should you do with that value? So for a new grad, you know, 739 00:33:25,906 --> 00:33:28,894 maybe it's just good enough that you, you know, you like being at 740 00:33:28,894 --> 00:33:30,304 work and that's like the mission. 741 00:33:30,605 --> 00:33:32,682 But I think as you get more senior, you should think about 742 00:33:32,682 --> 00:33:34,645 like really putting the mission first. 743 00:33:34,806 --> 00:33:37,853 And you know, so I think that we do do that. 744 00:33:37,853 --> 00:33:40,673 It's just the process might evolve year to year. 745 00:33:40,823 --> 00:33:44,486 And I will say that aside from mission for the engineering 746 00:33:44,486 --> 00:33:47,048 culture as well, like we've been, one of the things that 747 00:33:47,048 --> 00:33:48,857 we've been working on is evolving that culture more 748 00:33:48,857 --> 00:33:50,458 meaningfully, like more explicitly. 749 00:33:51,219 --> 00:33:54,866 If you worked for those of you who've like been at a big 750 00:33:54,866 --> 00:33:58,402 company, some companies may or may not have like a cultural 751 00:33:58,402 --> 00:34:01,425 statement in engineering or what it stands for, what good 752 00:34:01,425 --> 00:34:03,431 engineering looks like, the definition of good engineering. 753 00:34:03,861 --> 00:34:06,211 I mean, that's something that we've been working on because I 754 00:34:06,211 --> 00:34:07,931 want the team to understand that. 755 00:34:07,931 --> 00:34:11,556 And then, if the last part is like, how do you get the team to 756 00:34:11,556 --> 00:34:13,201 like operate under that? And that's like, you have to 757 00:34:13,201 --> 00:34:15,503 talk about it. You have to repeat it, you have 758 00:34:15,503 --> 00:34:17,879 to, to some extent, uh, you can incentivize it. 759 00:34:18,599 --> 00:34:21,043 I wanna be careful about, like, you can't, if you have a 760 00:34:21,043 --> 00:34:23,357 cultural statement of like be happy, you can't tell people to 761 00:34:23,357 --> 00:34:25,049 be happy and expect it to work out. 762 00:34:25,049 --> 00:34:28,079 So uh, you know, you get a low performance review for not being 763 00:34:28,079 --> 00:34:29,949 happy, right? So there's more to it than 764 00:34:29,949 --> 00:34:32,623 incentive structures in my mind. It's about living it, 765 00:34:32,623 --> 00:34:36,010 demonstrating it, and then elevating those who do that 766 00:34:36,010 --> 00:34:37,686 well. That's an incentive structure. 767 00:34:37,686 --> 00:34:40,716 But I don't wanna make it a stick for culture, at least. 768 00:34:41,471 --> 00:34:43,869 And so we do things like recognition and stuff for people 769 00:34:43,869 --> 00:34:47,072 who live those values. Yeah, and also probably also 770 00:34:47,072 --> 00:34:50,245 celebrate those kind of, uh, you know, values that people embody, 771 00:34:50,245 --> 00:34:51,871 right? So that maybe, you know, some 772 00:34:51,871 --> 00:34:53,129 examples that people can learn from. 773 00:34:53,609 --> 00:34:57,301 And I think I laugh when you say like, if your values is like, be 774 00:34:57,301 --> 00:35:00,074 happy, I don't know how you actually can assess someone 775 00:35:00,074 --> 00:35:01,814 happiness, uh, in your performance review. 776 00:35:02,444 --> 00:35:05,344 So, uh, maybe let's move on to the next topic which you have 777 00:35:05,344 --> 00:35:09,551 touched on a little bit as well. So in a pre recording of this 778 00:35:09,551 --> 00:35:12,905 conversation actually, you mentioned that you put a lot of 779 00:35:12,905 --> 00:35:15,893 emphasis in, you know, building management systems in your 780 00:35:15,893 --> 00:35:18,705 organization, right? So first of all, maybe tell us a 781 00:35:18,705 --> 00:35:21,527 little bit more about what is management system and why is it 782 00:35:21,527 --> 00:35:25,502 so important? So in a nutshell, it's about 783 00:35:25,502 --> 00:35:28,610 scaling repeatable outcomes. To give you an example of like a 784 00:35:28,610 --> 00:35:31,244 really rudimentary was, it's like to have a standup. 785 00:35:32,054 --> 00:35:34,754 Everyone understands that what a standup is, at least these days. 786 00:35:34,844 --> 00:35:36,344 You know, if you go back 20 years, maybe not. 787 00:35:36,674 --> 00:35:39,620 And like why does that exist? Well, it's a reliable thing that 788 00:35:39,620 --> 00:35:41,872 we can all look at and say, this happens. 789 00:35:41,902 --> 00:35:45,147 And that's the place that you tell your manager, hey, I'm 790 00:35:45,147 --> 00:35:47,468 stuck. Before standups existed, there 791 00:35:47,468 --> 00:35:50,051 was no form for that. So you might send an email and 792 00:35:50,051 --> 00:35:52,462 then you'd wait around. But that's not a system, because 793 00:35:52,462 --> 00:35:54,830 maybe your manager is different than other managers and they 794 00:35:54,830 --> 00:35:56,650 don't check their email until the end of the day. 795 00:35:56,650 --> 00:35:58,240 Now you're now a whole day goes by. 796 00:35:58,866 --> 00:36:02,489 And so someone created that notion of a standup to set an 797 00:36:02,489 --> 00:36:05,020 expectation to how to interact with your manager. 798 00:36:05,770 --> 00:36:08,676 And so, you know, Agile and stuff are systems like that, 799 00:36:08,676 --> 00:36:10,845 right? And so management systems are 800 00:36:10,845 --> 00:36:13,680 about how you can run a group of people. 801 00:36:14,357 --> 00:36:17,387 It's a set of processes and expectations that can scale. 802 00:36:17,777 --> 00:36:21,327 And key to that is that it's probably not stationary. 803 00:36:22,204 --> 00:36:25,191 And it falls with the company and the team but you wanna stand 804 00:36:25,191 --> 00:36:27,444 and formalize that. And so if you don't formalize 805 00:36:27,444 --> 00:36:31,118 it, then, you know, going back to the boxes, it's basically 806 00:36:31,118 --> 00:36:33,785 tribal. And so if the person that runs 807 00:36:33,785 --> 00:36:36,301 that thing disappears, you can lose that system. 808 00:36:36,881 --> 00:36:38,806 And so what you wanna do is you wanna make sure that as you 809 00:36:38,806 --> 00:36:42,070 scale, you codify these things. Much like a code base, right? 810 00:36:42,070 --> 00:36:44,440 Like you're writing code and you've got coding standards. 811 00:36:44,470 --> 00:36:47,134 Imagine trying to set coding standards without writing it 812 00:36:47,134 --> 00:36:49,455 down. But you just all sort of know 813 00:36:49,455 --> 00:36:50,680 what the coding standards are, right? 814 00:36:50,680 --> 00:36:53,930 Like you might get there, but then as the team grows, at some 815 00:36:53,930 --> 00:36:56,438 point, someone's doing really bad stuff and everyone is upset 816 00:36:56,438 --> 00:36:59,098 and no one knows why, right? And eventually you just lost 817 00:36:59,098 --> 00:37:00,610 that standard. So it's very similar. 818 00:37:00,610 --> 00:37:03,174 You want to think about like yes, you're doing certain 819 00:37:03,174 --> 00:37:05,434 activities. People probably understand what 820 00:37:05,434 --> 00:37:07,981 they are, but you're gonna benefit a lot if you can just 821 00:37:07,981 --> 00:37:11,280 write it down. Yeah, maybe some, give some, I 822 00:37:11,280 --> 00:37:13,899 dunno, tangible practice here. Because like coding standards, 823 00:37:13,899 --> 00:37:15,564 many engineers understand, you know. 824 00:37:15,714 --> 00:37:19,186 Probably you have, I dunno, the line width, tabs vs space, and 825 00:37:19,186 --> 00:37:21,234 you know, variable naming and things like that. 826 00:37:21,444 --> 00:37:24,534 How do you actually do that for management standards, right? 827 00:37:24,534 --> 00:37:26,874 So maybe give us some practical tips and examples. 828 00:37:26,934 --> 00:37:29,904 And how do you actually put it as a artifacts, right? 829 00:37:31,133 --> 00:37:33,745 Yeah. Like to give you like a really 830 00:37:33,745 --> 00:37:37,153 rudimentary example that we do, that we introduced is, you know, 831 00:37:37,153 --> 00:37:40,702 we do like a pretty standard process most companies do now 832 00:37:40,702 --> 00:37:43,582 where we do Scrum of Scrums. So Scrum is like a meeting. 833 00:37:43,702 --> 00:37:45,532 You have here the team, you talk about stuff. 834 00:37:45,922 --> 00:37:49,410 And the question is, one of the questions that's prevailing is 835 00:37:49,410 --> 00:37:53,544 when, how often, and in what forum to two teams that need to 836 00:37:53,544 --> 00:37:57,210 work together meet. And how do you keep, okay, but 837 00:37:57,210 --> 00:37:59,616 that, that times like 20 teams? How do you do that? 838 00:38:00,096 --> 00:38:06,424 And so a common solution to that is to do the review of projects 839 00:38:06,424 --> 00:38:08,916 that require interdependencies that are at conflict. 840 00:38:09,216 --> 00:38:12,472 In other words, Joey and Susie had to talk to each other to get 841 00:38:12,472 --> 00:38:14,268 something done. They couldn't quite resolve it 842 00:38:14,268 --> 00:38:16,176 for whatever reason. Probably good reason. 843 00:38:16,566 --> 00:38:20,166 Where is the forum to fix that or to resolve that? 844 00:38:20,376 --> 00:38:24,006 Or maybe Joey lost an engineer and needs help. 845 00:38:24,006 --> 00:38:26,753 Like what's the forum for that? So you create a standing forum 846 00:38:26,753 --> 00:38:28,003 for that. So that's an example. 847 00:38:28,423 --> 00:38:31,423 Um, another one that's like quite common is like, in 848 00:38:31,423 --> 00:38:34,555 engineering at least, is like an operational review where you 849 00:38:34,555 --> 00:38:38,799 look at the health of the systems with the goal of being 850 00:38:38,799 --> 00:38:42,373 proactive not reactive. And this gets into like yes, at 851 00:38:42,373 --> 00:38:45,313 the team level, you should be doing it, but you can do it at 852 00:38:45,313 --> 00:38:47,359 the organizational level, at the whole team level, at the very 853 00:38:47,359 --> 00:38:48,789 top of the organization in my view. 854 00:38:48,789 --> 00:38:51,761 "How are we doing?" That's a hard question to ask 855 00:38:51,761 --> 00:38:54,213 cause we've got dozens and dozens of systems so how do you, 856 00:38:54,213 --> 00:38:56,734 how do you know what's going on? And so you basically trickle 857 00:38:56,734 --> 00:38:59,319 that down and say, okay, I wanna see this data. 858 00:38:59,699 --> 00:39:01,379 Let's get that point down the org. 859 00:39:02,249 --> 00:39:05,425 You can see here though, a lot of this is about having good 860 00:39:05,425 --> 00:39:06,689 accounting. Uh, accounting is not the right 861 00:39:06,689 --> 00:39:08,823 word. But having good systems that 862 00:39:08,823 --> 00:39:12,084 track things, right? Whether it's work or it's 863 00:39:12,084 --> 00:39:14,700 metrics, data, whatever it might be. 864 00:39:15,522 --> 00:39:18,792 And I think, you know, there are standard HR things as well that 865 00:39:18,792 --> 00:39:21,772 evolved that maybe didn't exist at the beginning of the 866 00:39:21,772 --> 00:39:24,564 company's journey. Like there again, like you have 867 00:39:24,564 --> 00:39:27,671 no, non-existent HR when you're a team of three people and then 868 00:39:27,671 --> 00:39:30,553 you've got, you know, like the whole department of 10, 1,000 869 00:39:30,553 --> 00:39:32,515 people when you're like 100,000 employees. 870 00:39:32,965 --> 00:39:37,356 And somewhere in that journey, things like coaching plans get 871 00:39:37,356 --> 00:39:40,741 created or like performance management expectations. 872 00:39:40,741 --> 00:39:43,607 Like when we talk about performance management, so much 873 00:39:43,607 --> 00:39:47,413 of those expectations is about like the ICs feeling like, oh, 874 00:39:47,413 --> 00:39:49,959 great, now management team has this tool, this stick that 875 00:39:49,959 --> 00:39:52,681 they're gonna go and use on us and make us work harder. 876 00:39:52,801 --> 00:39:55,475 But that's actually a secondary point. 877 00:39:55,475 --> 00:39:58,685 Like there is an expectation on the managers to do that. 878 00:39:59,015 --> 00:40:00,935 And so there, this is again, the evolution. 879 00:40:00,935 --> 00:40:04,069 There are times where managers, I mean, we've all worked in a 880 00:40:04,069 --> 00:40:05,731 team where the manager wasn't doing their job. 881 00:40:06,454 --> 00:40:09,052 The person is bad. And for some reason that person 882 00:40:09,052 --> 00:40:11,329 is still here six months later. And you're, and you're sitting 883 00:40:11,329 --> 00:40:12,934 there losing respect for your manager every day. 884 00:40:12,934 --> 00:40:14,824 You're like why isn't this not person not doing their job? 885 00:40:15,184 --> 00:40:17,694 Well, I'll tell you why. It's like because there wasn't 886 00:40:17,694 --> 00:40:21,252 an expectation on that manager to hire good people, but also 887 00:40:21,252 --> 00:40:23,937 fire bad people. And hopefully, you're not gonna 888 00:40:23,937 --> 00:40:26,723 fire lots of people if you're like screening properly in the 889 00:40:26,723 --> 00:40:27,917 interview process. But it happens. 890 00:40:28,277 --> 00:40:31,247 And when that happens, I think there's an expectation on the 891 00:40:31,247 --> 00:40:32,909 manager. And so if the manager isn't 892 00:40:32,909 --> 00:40:35,576 doing that, then that's part of the system breaking down because 893 00:40:35,576 --> 00:40:38,687 you give all these tools and part of those tools is like, how 894 00:40:38,687 --> 00:40:42,019 do you use them? And so if the manager is like, I 895 00:40:42,019 --> 00:40:44,546 don't believe in firing people, not that they would say that, 896 00:40:44,546 --> 00:40:47,332 but let's just say, that would be an expectation that I think 897 00:40:47,332 --> 00:40:48,922 is an engineering leader I have to set. 898 00:40:48,952 --> 00:40:52,046 Because honestly, it's not fun doing that stuff. 899 00:40:52,046 --> 00:40:54,470 And so it's easy to say, I'm gonna focus more on building a 900 00:40:54,470 --> 00:40:58,862 great engineering system and focus less on having these hard 901 00:40:58,862 --> 00:41:01,915 conversations, right? And what I would do is as a CTO 902 00:41:01,915 --> 00:41:03,452 I would say, hey, we have to do that. 903 00:41:03,452 --> 00:41:06,632 It's managing the people is part of managing the systems because 904 00:41:06,632 --> 00:41:10,376 the people maintain the systems. And so you have to hold just as 905 00:41:10,376 --> 00:41:13,850 high of a bar on the people management part as you would on 906 00:41:13,850 --> 00:41:15,679 the system management. Yeah. 907 00:41:15,679 --> 00:41:18,325 So I think there are a few key things that I, uh, listen just 908 00:41:18,325 --> 00:41:20,743 now when you mention, you know, the explanation, right? 909 00:41:20,743 --> 00:41:23,876 So first thing is definitely expectations must be clear, 910 00:41:23,876 --> 00:41:25,751 right? Second thing is that you 911 00:41:25,751 --> 00:41:27,660 probably need to put it down somewhere, right? 912 00:41:27,660 --> 00:41:31,157 So whenever, you know, people leave, you should not have this 913 00:41:31,157 --> 00:41:33,436 knowledge also gone along with the person, right? 914 00:41:33,676 --> 00:41:35,986 So maybe the writing culture is very important as well. 915 00:41:35,986 --> 00:41:38,986 Maybe you can codify that as, I dunno, like how the way things 916 00:41:38,986 --> 00:41:41,116 work, way of working, some people call it, right? 917 00:41:41,446 --> 00:41:44,105 Or, you know, like GitLab also has a very good example where 918 00:41:44,105 --> 00:41:47,406 they codify everything and open source it, in fact for people to 919 00:41:47,406 --> 00:41:49,390 learn. So I think, uh, leadership also 920 00:41:49,390 --> 00:41:52,457 play a bigger, uh, role in terms of, you know, codifying these 921 00:41:52,457 --> 00:41:55,133 standards. So in terms of hiring managers, 922 00:41:55,133 --> 00:41:59,317 I think this is also part and parcel of like executive role. 923 00:41:59,467 --> 00:42:03,457 Because once you hire so-called bad managers, I think it's very, 924 00:42:03,457 --> 00:42:05,964 very difficult to actually maintain the standards, maintain 925 00:42:05,964 --> 00:42:07,837 the expectations, and things like that. 926 00:42:08,167 --> 00:42:11,211 So tell us, what's the importance of manager role and 927 00:42:11,211 --> 00:42:14,895 what do you think should executive do in order to avoid, 928 00:42:14,895 --> 00:42:16,957 you know, like hiring bad managers? 929 00:42:17,904 --> 00:42:20,784 Let me start with, um, the second half of that question, I 930 00:42:20,784 --> 00:42:23,502 think, which is like how, basically how to build a 931 00:42:23,502 --> 00:42:25,677 management team, right? I think that answer first. 932 00:42:25,677 --> 00:42:29,007 So for me, I think that you should be, we should be very 933 00:42:29,007 --> 00:42:33,507 clear with what managers do. And one of the most difficult 934 00:42:33,507 --> 00:42:37,979 things I think a startup faces is does a manager write code or 935 00:42:37,979 --> 00:42:40,957 not, for example. Because when you were a team of 936 00:42:40,957 --> 00:42:44,147 10 people and you were a CTO and you were managing five 937 00:42:44,147 --> 00:42:46,115 engineers, you were definitely writing code. 938 00:42:46,735 --> 00:42:49,395 At some point, that statement becomes false. 939 00:42:49,532 --> 00:42:52,457 And it becomes false because you're actually causing more 940 00:42:52,457 --> 00:42:55,112 problems as a manager writing code than helping. 941 00:42:55,112 --> 00:42:57,212 And it becomes... that's because there's all these other things 942 00:42:57,212 --> 00:42:59,114 you have to do that require your attention. 943 00:43:00,280 --> 00:43:01,540 Again, we have to evolve ourselves. 944 00:43:01,540 --> 00:43:04,759 And so one of the most important things to decide is when does 945 00:43:04,759 --> 00:43:07,393 that cutoff occur? When in the company's history do 946 00:43:07,393 --> 00:43:10,747 you stop asking managers to write code and manage them 947 00:43:10,747 --> 00:43:14,214 accordingly? And so, for example, I set very 948 00:43:14,214 --> 00:43:16,507 clear standards now in the company. 949 00:43:16,527 --> 00:43:18,517 I don't think there was a standard before I was a CTO. 950 00:43:18,517 --> 00:43:21,107 I think one of the first things I said is like, as a manager, 951 00:43:21,107 --> 00:43:22,764 you can write code. I mean, I don't, I'm not gonna 952 00:43:22,764 --> 00:43:26,809 stop you if you write code. But I will only allow you to be 953 00:43:26,809 --> 00:43:30,056 measured as a, in your performance by how you manage. 954 00:43:30,506 --> 00:43:33,218 So how great of code you've written does not factor into 955 00:43:33,218 --> 00:43:34,826 your performance review. You can write it. 956 00:43:35,394 --> 00:43:37,164 And if people like you for it, fantastic! 957 00:43:37,404 --> 00:43:39,669 But if you lose your best engineer to that because they're 958 00:43:39,669 --> 00:43:42,651 upset that you overruled them, I'm not gonna look at the cause, 959 00:43:42,651 --> 00:43:45,148 I'll look at the outcome. The outcome is that you lost 960 00:43:45,148 --> 00:43:48,149 your best engineer. And so setting that kind of 961 00:43:48,149 --> 00:43:50,290 requirement or expectation really crystallizes for managers 962 00:43:50,290 --> 00:43:53,249 what success looks like. Because I think the most 963 00:43:53,249 --> 00:43:56,298 important thing for engineering leaders, first and foremost, is 964 00:43:56,298 --> 00:43:59,501 that they can recruit good engineers, but also keep them, 965 00:43:59,501 --> 00:44:02,735 like empower them. So great engineers or great 966 00:44:02,735 --> 00:44:06,209 anything, any kinda employee, generally won't work with you if 967 00:44:06,209 --> 00:44:08,955 you will micromanage them and you're overruling them, right? 968 00:44:09,237 --> 00:44:11,629 Putting aside like the bad ones, the good ones for sure, they're 969 00:44:11,629 --> 00:44:13,853 not gonna want you telling them what to do cause you hire them 970 00:44:13,853 --> 00:44:17,654 to do that job. And so if you can't do that, you 971 00:44:17,654 --> 00:44:20,070 can't be successful, at least in a startup. 972 00:44:20,430 --> 00:44:21,570 Because how do you scale yourself? 973 00:44:21,570 --> 00:44:24,190 You have to hire a hundred engineers over the next two 974 00:44:24,190 --> 00:44:26,574 years as an organization and you're bleeding out only of the 975 00:44:26,574 --> 00:44:28,805 good ones. Like you're done as a company, 976 00:44:28,805 --> 00:44:30,618 right? You'll only keep the bad people 977 00:44:30,618 --> 00:44:32,280 who don't care that they're being micromanaged. 978 00:44:32,415 --> 00:44:35,152 But then you're also being limited because your systems are 979 00:44:35,152 --> 00:44:37,945 only as good as your managers who are writing code 30% of the 980 00:44:37,945 --> 00:44:41,274 time. So they should be technical so 981 00:44:41,274 --> 00:44:45,094 they can understand BS, right? But it has to be low ego, 982 00:44:45,094 --> 00:44:47,304 because like I said earlier, even if they were technical just 983 00:44:47,304 --> 00:44:50,310 two years ago, the thing that they were doing two years ago 984 00:44:50,310 --> 00:44:53,269 might be an antipattern now. I mean just think of like AI 985 00:44:53,269 --> 00:44:55,924 systems and like RAG and stuff. Like the patterns are changing 986 00:44:55,924 --> 00:44:58,658 so fast that what was state-of-the art two years ago 987 00:44:58,658 --> 00:45:00,259 is kind of like yesterday's news now. 988 00:45:00,649 --> 00:45:04,582 And so, so you have to have that mentality that I'm here 989 00:45:04,582 --> 00:45:07,829 ultimately to go and hire this person that I used to work with 990 00:45:07,829 --> 00:45:09,377 who's amazing and then get outta the way. 991 00:45:10,239 --> 00:45:12,969 And I always, I say like the it's the player coach model. 992 00:45:12,969 --> 00:45:16,041 Like I want managers to coach, but I don't want them over there 993 00:45:16,041 --> 00:45:18,516 scoring the goals. I mean, they're there, they're 994 00:45:18,516 --> 00:45:20,577 gonna help the team, but that's someone else's job. 995 00:45:21,257 --> 00:45:25,474 So I think if you could do that well then you can scale for a 996 00:45:25,474 --> 00:45:27,638 while. And I think it's true 997 00:45:27,638 --> 00:45:29,683 universally, frankly, in any good engineering org. 998 00:45:30,834 --> 00:45:33,945 Yeah, I think the art of letting go is always the difficult one, 999 00:45:33,945 --> 00:45:36,104 especially if you start as a good IC, right? 1000 00:45:36,104 --> 00:45:39,314 As you grow into senior, you develop solutions, write code, 1001 00:45:39,314 --> 00:45:42,254 and typically it's like great solutions, great code, right? 1002 00:45:42,344 --> 00:45:44,054 And then you move into the management role. 1003 00:45:44,054 --> 00:45:46,778 That's when actually the trade-off starts to become like 1004 00:45:46,778 --> 00:45:49,214 a battle, right? So do I write less code? 1005 00:45:49,214 --> 00:45:52,096 Do I, you know, hand it off to the engineers? 1006 00:45:52,382 --> 00:45:56,097 So maybe for the strong ICs out there who typically avoid, you 1007 00:45:56,097 --> 00:45:59,406 know, being a manager simply because they still love to write 1008 00:45:59,406 --> 00:46:02,056 some code, right? Do you think it's a good thing 1009 00:46:02,056 --> 00:46:03,784 that they should just focus on writing code? 1010 00:46:04,084 --> 00:46:07,752 Or they should maybe try being a manager or some kind of a leader 1011 00:46:07,752 --> 00:46:10,105 role? And how can they actually detach 1012 00:46:10,105 --> 00:46:12,487 from, you know, the necessity of writing code? 1013 00:46:13,513 --> 00:46:16,824 I usually, if someone has to ask me, I usually tell them, don't 1014 00:46:16,824 --> 00:46:20,081 go into management. It's an entirely different 1015 00:46:20,081 --> 00:46:22,681 skillset. You have way less control than 1016 00:46:22,681 --> 00:46:25,483 you think you do, because ultimately you're not the one 1017 00:46:25,483 --> 00:46:28,267 steering the wheel, right? Imagine trying to tell your Uber 1018 00:46:28,267 --> 00:46:29,995 driver where to go from the backseat. 1019 00:46:29,995 --> 00:46:32,095 Like that's what it's like, but you do for that team of people. 1020 00:46:32,575 --> 00:46:35,785 And so it can feel in some ways, like it looks like you have all 1021 00:46:35,785 --> 00:46:38,443 the authority, but you have none of the actual, like direct 1022 00:46:38,443 --> 00:46:40,985 control in some ways. And so it can be frustrating. 1023 00:46:40,985 --> 00:46:42,575 It's an entirely different skillset. 1024 00:46:42,769 --> 00:46:45,401 The lows are very low. Like when you're, you know, 1025 00:46:45,401 --> 00:46:48,291 politics, in a big company, the politics can crush you. 1026 00:46:48,936 --> 00:46:51,044 But you're also, in a small company, expected to write code 1027 00:46:51,044 --> 00:46:53,136 while manage, like it's a very challenging place. 1028 00:46:53,136 --> 00:46:55,086 You have to figure out the right balance to use your skillset. 1029 00:46:55,086 --> 00:46:58,822 So I always tell ICs, if you're capable of writing code, you 1030 00:46:58,822 --> 00:47:01,983 should stick with that for as long as you can until you feel 1031 00:47:01,983 --> 00:47:03,470 like that's not, you're not capable anymore. 1032 00:47:04,076 --> 00:47:07,621 Management is not, like a side hobby to try out. 1033 00:47:07,651 --> 00:47:10,208 It's cause if you think about it, if someone says, hey, Michi, 1034 00:47:10,228 --> 00:47:12,418 I wanna try managing for a little bit, what do you think? 1035 00:47:13,138 --> 00:47:16,148 I want you to imagine me saying, cool, Henry, this is your 1036 00:47:16,148 --> 00:47:18,598 manager now. They just, they just wanna try 1037 00:47:18,598 --> 00:47:20,713 it out, right? Like you're not cool with that. 1038 00:47:21,073 --> 00:47:23,743 So you need to hear from me, no, no, they're all in. 1039 00:47:24,133 --> 00:47:26,113 They're here. They're gonna be here for years. 1040 00:47:26,113 --> 00:47:27,823 They're gonna help you succeed, right? 1041 00:47:27,823 --> 00:47:30,953 So if that's not their mental model going into it, you're not 1042 00:47:30,953 --> 00:47:34,144 gonna, it's probably not gonna create good outcomes for you as 1043 00:47:34,144 --> 00:47:37,118 a report. And so I'm always, I'm like 1044 00:47:37,118 --> 00:47:40,610 really focused on the person having the right motives. 1045 00:47:40,670 --> 00:47:42,950 You know, really, they're not there for themselves. 1046 00:47:42,950 --> 00:47:44,720 They want to succeed for the business. 1047 00:47:44,720 --> 00:47:46,160 They wanna to help the team succeed. 1048 00:47:46,160 --> 00:47:48,650 They want people to succeed that are working with them. 1049 00:47:49,487 --> 00:47:51,951 And not because they want more authority over like what gets 1050 00:47:51,951 --> 00:47:53,549 done, right? Because the reality is it's not, 1051 00:47:53,549 --> 00:47:56,701 it's just not true. And some of the most difficult 1052 00:47:56,701 --> 00:48:00,683 parts of running a company is frankly like letting go of 1053 00:48:00,683 --> 00:48:03,613 people or changing projects. Like imagine you are working on 1054 00:48:03,613 --> 00:48:06,573 this amazing new system for six months, and then I tell you 1055 00:48:06,573 --> 00:48:09,197 we're not gonna do that anymore. Like that's a really hard 1056 00:48:09,197 --> 00:48:11,202 conversation, because people might quit over that kind of 1057 00:48:11,202 --> 00:48:13,397 stuff. Or I could say, you're not the 1058 00:48:13,397 --> 00:48:14,807 right person to maintain that anymore. 1059 00:48:15,367 --> 00:48:17,227 Those conversations have to be done by the manager. 1060 00:48:17,227 --> 00:48:21,925 And if that stuff, firing people and like doing HR stuff is not 1061 00:48:21,925 --> 00:48:24,244 something that you think, I mean, not that anyone enjoys 1062 00:48:24,244 --> 00:48:26,725 that, but if that's not something that you think would 1063 00:48:26,725 --> 00:48:29,774 be something you would want to get better at, it's not the 1064 00:48:29,774 --> 00:48:32,368 right career for you. So I always tell people you need 1065 00:48:32,368 --> 00:48:35,224 to have the right motivations and really be in it to win it to 1066 00:48:35,224 --> 00:48:38,102 be a manager. Cause it's just not a path that 1067 00:48:38,102 --> 00:48:40,622 is necessarily that it's not all fun and games. 1068 00:48:40,982 --> 00:48:42,812 At the same time, it's also a one way path. 1069 00:48:42,962 --> 00:48:44,782 Like I think it's very difficult to go back. 1070 00:48:45,422 --> 00:48:48,281 So once you step into this thing, like I said, you have to 1071 00:48:48,281 --> 00:48:50,697 be committed to it. It takes a couple of years just 1072 00:48:50,697 --> 00:48:52,452 like with engineering to get good at it. 1073 00:48:53,032 --> 00:48:55,090 You know, no one gets to go and say, oh, I'm gonna just hobby 1074 00:48:55,090 --> 00:48:56,642 code for a little bit, try out the career. 1075 00:48:56,642 --> 00:48:58,382 Like, no, it's gonna take you years to get good at it. 1076 00:48:58,382 --> 00:49:00,386 It's the same with managing. It takes you years to get good 1077 00:49:00,386 --> 00:49:02,522 at it. And so by the time you're like, 1078 00:49:02,522 --> 00:49:05,072 okay, I kinda understand what's going on here, but maybe this 1079 00:49:05,072 --> 00:49:07,669 isn't for me. It's a year or two has gone by, 1080 00:49:07,669 --> 00:49:09,184 right? Because you have to go through 1081 00:49:09,184 --> 00:49:11,464 at least one promotion cycle to see what it's like to promote 1082 00:49:11,464 --> 00:49:12,359 people. That takes some time. 1083 00:49:12,979 --> 00:49:16,119 So a year's a long time to not be in the code. 1084 00:49:16,119 --> 00:49:19,695 And while you can go back, it could probably set you back. 1085 00:49:20,175 --> 00:49:23,063 And so I always say like just it's not a step that I would 1086 00:49:23,063 --> 00:49:24,355 take lightly. Something that you should really 1087 00:49:24,355 --> 00:49:27,057 think about. If you really enjoy 1088 00:49:27,057 --> 00:49:30,584 coordinating, helping people, empowering people, hiring, 1089 00:49:30,584 --> 00:49:32,786 right? I think it's something worth 1090 00:49:32,786 --> 00:49:35,523 taking a look at. But just know that just like 1091 00:49:35,523 --> 00:49:37,245 with any job it has its downsides. 1092 00:49:37,835 --> 00:49:40,435 You know, I would say like the worst part about being an IC is 1093 00:49:40,435 --> 00:49:43,572 probably fixing bugs and doing on call, fires and incidents, 1094 00:49:43,572 --> 00:49:45,010 right? Like that's probably the worst 1095 00:49:45,010 --> 00:49:46,475 part. And if someone said to you, 1096 00:49:46,475 --> 00:49:48,385 yeah, but I don't, I don't wanna do that stuff. 1097 00:49:48,385 --> 00:49:51,070 Like I just like writing code. You would think that they're 1098 00:49:51,070 --> 00:49:52,475 quite naive as an engineer, right? 1099 00:49:52,475 --> 00:49:56,767 And so you shouldn't love waking up at 2:00 AM to fix a fire on a 1100 00:49:56,767 --> 00:49:57,905 system. No one does. 1101 00:49:58,295 --> 00:49:59,855 But it's something that you might get good at. 1102 00:49:59,860 --> 00:50:02,045 You know, you get really good at like handling an incident. 1103 00:50:02,465 --> 00:50:04,709 And you know it's a skill set that you have to get really good 1104 00:50:04,709 --> 00:50:06,993 at to become senior. And you have to know, you know, 1105 00:50:06,993 --> 00:50:09,179 there's gonna be some fire and you have to fix it very quickly 1106 00:50:09,179 --> 00:50:11,994 under a lot of pressure. That is part of the job for 1107 00:50:11,994 --> 00:50:14,145 being an IC. There's an equivalent of that in 1108 00:50:14,145 --> 00:50:16,530 managers and if that stuff is like terrifying to you, you're 1109 00:50:16,530 --> 00:50:18,892 like, I don't wanna do that. Just I would think twice. 1110 00:50:19,858 --> 00:50:22,552 Yeah, dealing with people is always the, maybe the art, 1111 00:50:22,552 --> 00:50:24,085 right, when become a manager, right? 1112 00:50:24,085 --> 00:50:28,465 So because like code can be very structured and maybe it's more 1113 00:50:28,465 --> 00:50:30,637 like binary, right? One or no, yes or no, right? 1114 00:50:30,637 --> 00:50:33,431 But with people, sometimes it's very ambiguous and like what you 1115 00:50:33,431 --> 00:50:35,841 said, right? You cannot put a control on, you 1116 00:50:35,841 --> 00:50:38,517 know, like you think you are the leader and you have the 1117 00:50:38,517 --> 00:50:40,243 authority, right? But sometimes you actually 1118 00:50:40,243 --> 00:50:42,581 cannot control what actually happened in reality. 1119 00:50:42,817 --> 00:50:45,351 So I think, uh, yeah, being a manager definitely requires a 1120 00:50:45,351 --> 00:50:47,861 different skill set and sometimes it can be really 1121 00:50:47,861 --> 00:50:50,291 rewarding becoming a good manager that people love, right? 1122 00:50:50,291 --> 00:50:52,778 And achieving a certain outcomes as a team, right? 1123 00:50:52,977 --> 00:50:55,129 But definitely it requires a different skill set. 1124 00:50:55,583 --> 00:50:58,257 So how about becoming a CTO? Because I understand that 1125 00:50:58,257 --> 00:51:00,928 becoming a manager is one thing, becoming a, you know, like a 1126 00:51:00,928 --> 00:51:03,428 people leader is one thing. But becoming an executive, I 1127 00:51:03,428 --> 00:51:06,534 dunno, like maybe in your role, right, currently in the Garner 1128 00:51:06,534 --> 00:51:09,558 Health as a CTO, I think it requires a different skillset as 1129 00:51:09,558 --> 00:51:11,264 well. So what do you think are the 1130 00:51:11,264 --> 00:51:13,922 difference between, you know, manager, people manager, and 1131 00:51:13,922 --> 00:51:16,900 also the CTO role as a executive role? 1132 00:51:17,968 --> 00:51:20,667 I dunno if you're, I'm not sure if there's two, but there's at 1133 00:51:20,667 --> 00:51:22,351 least one I can think of. Let me start with the first one 1134 00:51:22,351 --> 00:51:23,761 I think of, and then I'll get to the second. 1135 00:51:23,971 --> 00:51:26,731 So the first one is maybe not necessarily exclusive to CTO, 1136 00:51:26,731 --> 00:51:30,106 but certainly at the more senior levels, like you have to be 1137 00:51:30,106 --> 00:51:32,381 proactive, not reactive. You have to be thinking about 1138 00:51:32,381 --> 00:51:34,960 what's happening in six to 12 months or 18 months, or the 1139 00:51:34,960 --> 00:51:36,943 bigger the company, the further I had to be thinking. 1140 00:51:37,363 --> 00:51:41,921 And if you're reacting, then you're, you're screwed. 1141 00:51:41,921 --> 00:51:44,663 And the reason you're screwed is because, going back to the Uber 1142 00:51:44,663 --> 00:51:48,601 driver thing, imagine that I'm managing an Uber driver and then 1143 00:51:48,601 --> 00:51:51,243 imagine that there's someone managing me on how to drive that 1144 00:51:51,243 --> 00:51:53,621 Uber driver. Like there's so many abstraction 1145 00:51:53,621 --> 00:51:55,896 layers and hops. And now imagine that's not 1146 00:51:55,896 --> 00:51:59,129 happening in real time. And so by the time I realize, 1147 00:51:59,129 --> 00:52:03,453 oh, I need to react, the time it takes for the organization to 1148 00:52:03,453 --> 00:52:07,796 shift course and adjust to that reaction is, at best case, 1149 00:52:07,796 --> 00:52:10,431 weeks, right? And worst case it could never 1150 00:52:10,431 --> 00:52:11,455 happen. They're just like, nope, we're 1151 00:52:11,455 --> 00:52:13,347 not doing that. But it's probably months or 1152 00:52:13,347 --> 00:52:15,352 quarters and that could kill you as a company. 1153 00:52:15,852 --> 00:52:18,942 So that statement of being proactive is true probably for 1154 00:52:18,942 --> 00:52:21,572 any organizational size, just because the higher up you go the 1155 00:52:21,572 --> 00:52:24,646 more, you have to account for the fact that the team will ship 1156 00:52:24,646 --> 00:52:27,588 slowly, right? But I think the second one is 1157 00:52:27,588 --> 00:52:31,458 that's unique mostly to CTOs is that you are the final line 1158 00:52:31,458 --> 00:52:33,810 between the engineering and the business. 1159 00:52:33,810 --> 00:52:38,202 Like there is no one above me to communicate to the business what 1160 00:52:38,202 --> 00:52:40,712 engineering is doing. Whereas there's always someone 1161 00:52:40,712 --> 00:52:42,518 above you, in any other role, right? 1162 00:52:42,518 --> 00:52:45,547 So if you're dealing with a business stakeholder who doesn't 1163 00:52:45,547 --> 00:52:48,497 quite get it and you're trying to say that stuff to them, maybe 1164 00:52:48,497 --> 00:52:50,705 your manager's got your back and they'll figure out a way to say 1165 00:52:50,705 --> 00:52:52,688 it, right? But it's, that chain stops. 1166 00:52:52,718 --> 00:52:54,558 And it finally at the CTO, it stops. 1167 00:52:54,558 --> 00:53:00,008 And it's just you and everybody else has specialization in their 1168 00:53:00,008 --> 00:53:03,501 stuff and very likely none of it's in engineering or very 1169 00:53:03,501 --> 00:53:05,966 little. And so be able to communicate 1170 00:53:05,966 --> 00:53:07,318 that. It's a bidirectional 1171 00:53:07,318 --> 00:53:08,479 communication, by the way, right? 1172 00:53:08,479 --> 00:53:11,719 Like you have to be able to communicate with the legal team. 1173 00:53:11,719 --> 00:53:13,849 And there are things like the hiring contracts. 1174 00:53:13,849 --> 00:53:16,691 And then there's like, I dunno if you wanna do open source and 1175 00:53:16,691 --> 00:53:19,219 there's like some license, on those topics. 1176 00:53:19,219 --> 00:53:22,537 And then with the sales team and then, you know, marketing and 1177 00:53:22,537 --> 00:53:24,929 people team. So you're the final arbiter on 1178 00:53:24,929 --> 00:53:27,882 this stuff. And I think, so the meta point 1179 00:53:27,882 --> 00:53:30,609 to all that is that communication is probably the 1180 00:53:30,609 --> 00:53:33,719 most important skill set because you have to communicate 1181 00:53:33,719 --> 00:53:36,999 something that a lot of atomic engineering decisions far down 1182 00:53:36,999 --> 00:53:40,499 the org and to translate that all the way up to the highest 1183 00:53:40,499 --> 00:53:43,117 levels of the company to people that are probably not technical 1184 00:53:43,117 --> 00:53:45,260 at all. And they expect you to be that 1185 00:53:45,260 --> 00:53:47,951 translator for them. And then vice versa, right? 1186 00:53:47,971 --> 00:53:49,969 'Cause they're like, hey, there's this compliance thing or 1187 00:53:49,969 --> 00:53:51,301 whatever. Like if it doesn't get 1188 00:53:51,301 --> 00:53:54,343 translated back, it gets lost. And this gets into like 1189 00:53:54,343 --> 00:53:56,941 management systems again. Like how do you route this work? 1190 00:53:56,941 --> 00:53:58,951 And it's a product management question as well. 1191 00:53:58,951 --> 00:54:01,674 There's other partners I might work with, but there is no one 1192 00:54:01,674 --> 00:54:03,804 else that I can rely on. It falls on me. 1193 00:54:03,864 --> 00:54:06,234 And if I'm lucky, my stakeholders are technical. 1194 00:54:06,534 --> 00:54:08,498 But you can't rely on that to be true. 1195 00:54:09,294 --> 00:54:11,604 Yeah, and sometimes also you need to handle complaints from 1196 00:54:11,604 --> 00:54:14,184 other departments as well. Like for example, if let's say 1197 00:54:14,184 --> 00:54:15,939 engineering are not delivering, engineering always having 1198 00:54:15,939 --> 00:54:18,564 incidents and things like that. So I think, yeah, you're kind of 1199 00:54:18,564 --> 00:54:20,574 like the bridge between the business and the engineering. 1200 00:54:20,604 --> 00:54:23,574 And I think the important thing is the communication, uh, also 1201 00:54:23,574 --> 00:54:25,979 understanding systems, right, management systems, and that 1202 00:54:25,979 --> 00:54:27,754 things are becoming more important. 1203 00:54:27,794 --> 00:54:29,323 And also designing those systems, right? 1204 00:54:29,323 --> 00:54:33,616 Like if you're below CTO, they're probably setting those 1205 00:54:33,616 --> 00:54:36,025 management systems. But the accountability for 1206 00:54:36,025 --> 00:54:38,468 creating those systems, it falls on me. 1207 00:54:38,828 --> 00:54:42,248 I'll also say that like the accountability to know which 1208 00:54:42,248 --> 00:54:46,264 systems you need and be proactive about that is also on 1209 00:54:46,264 --> 00:54:48,272 me. And so again, it goes back to 1210 00:54:48,272 --> 00:54:51,305 like, you just have to, as a, this is true for any executive 1211 00:54:51,305 --> 00:54:53,540 role, right? You just have to think about 1212 00:54:53,540 --> 00:54:56,925 what level of scale you're at. When is the right time to 1213 00:54:56,925 --> 00:54:58,995 introduce something that's annoying that, you know, might 1214 00:54:58,995 --> 00:55:01,475 feel like busy work for people, but it's actually really 1215 00:55:01,475 --> 00:55:05,379 important to scale the org. I think a good example debate we 1216 00:55:05,379 --> 00:55:07,845 have is metrics. Metrics is actually a great one 1217 00:55:07,845 --> 00:55:10,371 because some companies start with metrics when they're very 1218 00:55:10,371 --> 00:55:12,181 young. But realistically where you're 1219 00:55:12,181 --> 00:55:15,459 more interested in closing the sale than looking at a 1220 00:55:15,459 --> 00:55:17,381 dashboard. And so you're just writing the 1221 00:55:17,381 --> 00:55:19,292 thing. At some point, some of the very 1222 00:55:19,292 --> 00:55:22,052 low level metrics, these systems becomes important to look at it 1223 00:55:22,052 --> 00:55:24,857 at an aggregate, but may not matter in the moment when you're 1224 00:55:24,857 --> 00:55:26,522 building it, when you're just shipping it for money. 1225 00:55:27,251 --> 00:55:29,420 But if you go to like really, really big companies, everything 1226 00:55:29,420 --> 00:55:31,235 is measured out. Everything is every, there's a 1227 00:55:31,235 --> 00:55:32,481 KPI of everything. Every, every... 1228 00:55:32,481 --> 00:55:35,227 I mean there's different levels of maturity on this, but 1229 00:55:35,227 --> 00:55:37,475 generally the businesses that are much more data driven, which 1230 00:55:37,475 --> 00:55:40,276 sort really, really big. So somewhere between when you're 1231 00:55:40,276 --> 00:55:43,941 a two person startup and when you're at a 20,000 person 1232 00:55:43,941 --> 00:55:47,123 company, you've brought in this concept and there's different 1233 00:55:47,123 --> 00:55:50,118 levels of that. At an extreme, you would be 1234 00:55:50,118 --> 00:55:52,056 doing like, at scale experimentation, A/B testing. 1235 00:55:52,896 --> 00:55:54,216 So the question is, when do you bring that in? 1236 00:55:54,216 --> 00:55:56,779 So those kinds of things, I think are the things that, 1237 00:55:56,779 --> 00:55:59,283 again, it doesn't have to all come from me, but I'm 1238 00:55:59,283 --> 00:56:02,368 accountable for it. Yeah, so thanks for mentioning 1239 00:56:02,368 --> 00:56:04,083 that, how do you create systems, right? 1240 00:56:04,083 --> 00:56:06,543 I think it's very important as an executive to actually first 1241 00:56:06,543 --> 00:56:09,263 look beyond, right? Uh, your job is to be proactive, 1242 00:56:09,263 --> 00:56:11,849 not reactive, right? And how to design systems to, in 1243 00:56:11,849 --> 00:56:14,174 order to be more proactive towards those kind of things. 1244 00:56:14,594 --> 00:56:16,664 So you have worked in multiple companies, right? 1245 00:56:16,814 --> 00:56:19,544 You know, Stripe, Meta, Capital One and all that. 1246 00:56:19,814 --> 00:56:22,187 Do you think there are some non-intuitive thing that you 1247 00:56:22,187 --> 00:56:24,665 learned from those organizations that you wanna share with us as 1248 00:56:24,665 --> 00:56:27,824 well? I, those are very different 1249 00:56:27,824 --> 00:56:29,144 companies with different philosophies. 1250 00:56:29,144 --> 00:56:31,267 You wanna try to clarify a little bit more what you wanna 1251 00:56:31,267 --> 00:56:32,405 learn? So something that is 1252 00:56:32,405 --> 00:56:34,587 non-intuitive that you think, oh, this is very unique from 1253 00:56:34,587 --> 00:56:37,537 this company that I learned that I think could be useful if you 1254 00:56:37,537 --> 00:56:41,477 wanna share with others. So I spent a big chunk of my 1255 00:56:41,477 --> 00:56:43,802 career doing payment stuff. And I think one thing that's 1256 00:56:43,802 --> 00:56:47,361 interesting about the financial industry is that they're like 1257 00:56:47,361 --> 00:56:49,563 excessively obsessive about fraud. 1258 00:56:50,267 --> 00:56:53,085 I would say like if you're, if it's a fraud topic, you get 1259 00:56:53,085 --> 00:56:55,988 infinity budget. That's probably the biggest sort 1260 00:56:55,988 --> 00:56:58,109 of delta between that industry and others. 1261 00:56:58,669 --> 00:57:01,779 Even though it's funny 'cause fraud represents money loss, but 1262 00:57:01,779 --> 00:57:04,843 not money gain. So it's a very tricky balance 1263 00:57:04,843 --> 00:57:07,147 because you're trying to build features, but there's new 1264 00:57:07,147 --> 00:57:09,519 features all have some kind of fraud factor. 1265 00:57:10,399 --> 00:57:14,332 So there's like this internal war always between sort of like 1266 00:57:14,332 --> 00:57:18,275 the compliance and control and risk side, security side of the 1267 00:57:18,275 --> 00:57:21,148 business with the sort of the revenue and feature generation 1268 00:57:21,148 --> 00:57:25,432 part of the business. And getting that right is really 1269 00:57:25,432 --> 00:57:27,941 critical. I got, I used to, I did 1270 00:57:27,941 --> 00:57:30,913 consulting once for I don't want to name the name, but at a 1271 00:57:30,913 --> 00:57:31,943 financial institution, it's not Capital One. 1272 00:57:32,363 --> 00:57:34,307 And it was an innovation project. 1273 00:57:34,307 --> 00:57:35,987 They were working on building this new thing. 1274 00:57:36,407 --> 00:57:38,824 It was gonna be this amazing new product, consumer product. 1275 00:57:38,914 --> 00:57:42,880 And this is like a global brand. And, internally it was a huge 1276 00:57:42,880 --> 00:57:46,260 fight because every other department was like this is not 1277 00:57:46,260 --> 00:57:48,602 worth innovating on. It's gonna disrupt our business 1278 00:57:48,602 --> 00:57:50,688 model. And what about all these 1279 00:57:50,688 --> 00:57:52,836 different ways of. And basically they use fraud as 1280 00:57:52,836 --> 00:57:54,852 like risk as a way to kill the project. 1281 00:57:54,852 --> 00:57:58,017 And so, uh, I think they were talking about like phishing and, 1282 00:57:58,677 --> 00:58:00,532 you know, the login screen can have phishing attempts and what. 1283 00:58:00,997 --> 00:58:04,093 And so it's like what? I would say that they were 1284 00:58:04,093 --> 00:58:07,246 probably valid, I don't know. But, you know, the innovation is 1285 00:58:07,246 --> 00:58:10,102 very difficult in that space because you have this like huge 1286 00:58:10,102 --> 00:58:12,840 department in the company who's motivated to just kill any good 1287 00:58:12,840 --> 00:58:14,625 idea, and for good reason perhaps. 1288 00:58:14,625 --> 00:58:15,915 And the company's gonna figure that out. 1289 00:58:15,945 --> 00:58:18,663 Like Square and Stripe have figured out that balance 1290 00:58:18,663 --> 00:58:20,918 perhaps, but it's, um, interesting to see that war. 1291 00:58:20,978 --> 00:58:21,878 So that was something I observed. 1292 00:58:22,655 --> 00:58:25,621 Yeah, thanks for sharing that. And it's always, uh, you know, 1293 00:58:25,621 --> 00:58:28,805 challenging to prevent fraud and maybe if I can bring it up also, 1294 00:58:28,805 --> 00:58:30,295 security, right? It's always a balance, right? 1295 00:58:30,295 --> 00:58:32,781 You wanna be secure, but at the same time, you want to innovate 1296 00:58:32,931 --> 00:58:35,846 and create new things and disrupt the market and all that, 1297 00:58:35,846 --> 00:58:37,191 right? But sometimes security also 1298 00:58:37,191 --> 00:58:40,528 needs to take a, you know, priority as well to protect your 1299 00:58:40,528 --> 00:58:42,958 users, protect your systems, and protect your company altogether. 1300 00:58:43,438 --> 00:58:45,748 So, Michi, I think, uh, we learned a lot from the 1301 00:58:45,748 --> 00:58:47,444 conversation. Before I let you go, right, as 1302 00:58:47,444 --> 00:58:50,188 we wrap up our conversation, I would like to ask you one more 1303 00:58:50,188 --> 00:58:52,251 question. Typically, I ask the question 1304 00:58:52,251 --> 00:58:54,136 called the three technical leadership wisdom. 1305 00:58:54,156 --> 00:58:56,184 If you can think of it just like three advice that you want to 1306 00:58:56,184 --> 00:58:58,836 give to the listeners, what would your advice be? 1307 00:58:59,904 --> 00:59:02,260 Yeah. Uh, the first one I'd say is be 1308 00:59:02,260 --> 00:59:04,194 willing to be wrong as I said earlier. 1309 00:59:04,194 --> 00:59:06,894 Like you have to, if you don't do it, no one else will. 1310 00:59:07,680 --> 00:59:11,130 And be willing to make very hard decisions as a leader. 1311 00:59:11,549 --> 00:59:15,248 Choice A versus B. Not making a decision is in 1312 00:59:15,248 --> 00:59:18,513 itself making a decision. And so I'm sure you can all 1313 00:59:18,513 --> 00:59:20,920 imagine those times where you had some proposals for an 1314 00:59:20,920 --> 00:59:22,971 executive and then you're all waiting around for them to make 1315 00:59:22,971 --> 00:59:24,093 a decision. Like that's a decision. 1316 00:59:24,093 --> 00:59:25,363 You've just killed the team for a week. 1317 00:59:25,993 --> 00:59:30,102 Um, and finally I'd say, there's a famous sort of adage from, uh, 1318 00:59:30,102 --> 00:59:33,070 the CTO of Meta. But he used to say that 1319 00:59:33,070 --> 00:59:36,090 communication is the job. And I'm a big believer in that 1320 00:59:36,090 --> 00:59:37,966 as well. Like I think if you can't 1321 00:59:37,966 --> 00:59:40,189 communicate well in various ways, not just what I'm 1322 00:59:40,189 --> 00:59:42,194 thinking, but in terms of what the priorities are, what people 1323 00:59:42,194 --> 00:59:43,876 should be doing, what the expectations are. 1324 00:59:44,086 --> 00:59:46,918 If you can't do those things well, then you probably can't be 1325 00:59:46,918 --> 00:59:49,842 successful as a leader. Well, lovely wisdom. 1326 00:59:49,892 --> 00:59:51,152 So thanks for sharing them, right. 1327 00:59:51,152 --> 00:59:53,494 Especially the last part, right? Communication is definitely the 1328 00:59:53,494 --> 00:59:56,429 job of leader, right? Uh, especially if you wanna 1329 00:59:56,429 --> 00:59:58,772 communicate, over-communicate and make sure that the teams are 1330 00:59:58,772 --> 01:00:00,092 aligned and get the priorities right. 1331 01:00:00,392 --> 01:00:02,602 So Michi, if people love this conversation, they wanna connect 1332 01:00:02,602 --> 01:00:04,492 with you, they wanna find you online, is there place where 1333 01:00:04,492 --> 01:00:08,468 they can reach out to you? I guess, uh, if you're more 1334 01:00:08,468 --> 01:00:10,796 interested in Garner, take a look at getgarner.com. 1335 01:00:11,318 --> 01:00:12,905 Otherwise, you can find me on LinkedIn. 1336 01:00:13,127 --> 01:00:14,659 It's Michi Kono. Right? 1337 01:00:14,659 --> 01:00:16,189 Thank you so much for your time today, Michi. 1338 01:00:16,189 --> 01:00:18,559 I think we all learn a lot of things, you know, your insights, 1339 01:00:18,559 --> 01:00:21,060 your wisdom, especially in becoming a better CTO and 1340 01:00:21,060 --> 01:00:22,574 technical leader. So thank you for that. 1341 01:00:23,380 --> 01:00:23,800 Thank you.