How to Achieve Predictable Software Delivery at Scale | Syngenta’s Jason Krohn

0:00By any objective measure of changes of code, If anything, like this one person looks 10, 15 times more productive than the other. But it doesn't mean that that's what's happening. Until we come up with an objective measure of value delivered somehow, there's no real way. And that's why we focus so heavily on teams, effectively. If anything, the value I find in the developer-level metrics is where do we need more support.

0:32costly DORA metrics, insights into the health of your engineering team don't have to be complicated or expensive. That's why Linear B is introducing free DORA metrics for all. Say goodbye to spreadsheets and manual tracking or paying for your DORA metrics. Linear B is giving away a free, comprehensive DORA dashboard packed with essential insights, including all four key DORA metrics tailored to your team's data, industry standard benchmarks for gauging performance and setting data-driven goals, plus additional leading metrics including merge frequency and pull request size. Empower your team with the metrics they deserve.

1:06Sign up for your free Dora dashboard today at linearb.io slash Dora, or follow the link in the show notes. Welcome back to Dev Interrupted, everyone. I'm your co-host, Connor Bronston, and today we're joined by Jason Krohn. Jason is the global head of delivery and digital product engineering at Syngenta. Jason, welcome to Dev Interrupted. Hey, listeners, a quick heads up that we're taking a week off next week, so we won't be dropping a new episode on Tuesday, March 5th. Instead, season four will resume on March 12th. Now, back to this week's episode. Hi, thank you for having me. Absolutely. It's fantastic to have you on.

1:38We've known each other for a year or two now, and you've got over a decade of experience leading teams around the world. And you've also done that while scaling your teams around the world. That experience has given you a really well-rounded perspective, and I think you're ahead of where a lot of other organizations want to be as far as their engineering metrics program, your culture impact, and how you approach it. So a couple of years ago, when we were having guests on the show, all I didn't want to talk about was that hyperscaling piece. But you've had to do it efficiently. It's not growth at all costs anymore.

2:08And you've really cared about culture throughout the entire process, about the experience your engineers are having, all while scaling from 150 to 400 developers. And so what I want to talk about is how do you scale efficiently? Considering the complexities that you faced, how do you build and grow teams while ensuring people are empowered and autonomous? What's your approach? So it starts with the people themselves, right? You have to hire the right people and then empower them with the things that they care about effectively, right? As we started small, it was easy to do, right? You have a small team.

2:42A small team is attached to a product. They get a thing that they can directly care about. It's very easy for them to feel engaged effectively, right? As your teams grow, though, you start to lose a lot of this. You start to bring in a little more of the enterprise. you start to bring in a little more of the friction to their work. And so as we've scaled, we needed to change a few of those things around effectively. So how do you identify what the right things are? What developers want, as you said? An important part of building and scaling a team for us, at least at Syngenta, is retention. So we want to hire good people, we want to get out of their way, but then we want to keep them.

3:21Your most impactful teams are the teams that have had people there for a while. Right. Who understand what's going on, who understand the complexities of what they're doing, who can then sit and partner with the business to really understand, hey, we have an opportunity. Can we seize it? What does it take? How do we do it? What's the right way to go about it? Right. So they have the cultural context and kind of the depth of knowledge that you want to see to lead teams, frankly. And the systems knowledge, right? So agriculture is very seasonal. And so when we come in with a market opportunity, the business side is going to want a thing done, right?

3:55That thing may not always be feasible. But if you have a team that deeply understands the product, the context of the business, the context of our user base, we can get to how can we achieve that value in the timeframe that we have effectively, right? Which may not be what is explicitly being asked for. Interesting. So that's how you are driving decisions that align with organizational goals, it sounds like. Absolutely, but you have to keep people for that, right? And for us, I'm a big believer in four pillars for people, right? So you have to compensate people fairly. You have to give them a pleasant place to work, give them work they care about, and give them growth, right?

4:36If you give them those four things, they will stay for the most part. There's always some people who they want to switch domains, they want to go do something else, and that's fine. But if you give people those four things, right, one, you can stand up and say, I value you and I respect you. And you're true to your word, effectively, right? If you don't give them any of those four things, though, they will go somewhere else to find those things. And so we really try hard as we scale to always keep those things in mind, right? To from a salary perspective, we need to be competitive. I can't say that I value what you bring to our organization and I care about you and underpay you.

5:15Right? Now, that doesn't mean we have to be top of the market for everything all the time, but I have to pay you fairly for what you bring effectively. Right? We have to give you a pleasant place to work. And that can mean everything from your laptop can't be slow for what you do. The build processes can't be onerous, right? You can't have all these little annoyances during your day that make you not like your job effectively, but it's also our office, the culture that we have. You want coming to work to be pleasant, not something you don't like effectively, right? That makes total sense. And it aligns to so much research in the field of happy teams that feel engaged, like they can make an impact, are more productive, they stay longer.

6:00This is very clearly how you're building this organizational matrix, how you're rewiring the social circuitry of your organization to work and retain folks. But what are the kind of guidelines you put into place to make that happen. For the rest of that retention. Yeah. Because then there's the other two dimensions. Those are two. The other two is I got to give you work you care about. Thankfully, it's in Genzo. We have work people can care about, right? We're all about fighting climate change, sustainable agriculture, regenerative agriculture. How do we use technology and precision agriculture machinery to help people do better, to do more?

6:35That's better for them and better for the earth. So it's thankfully been easy to get people on board and the things around that. There's data science around that. There's systems work. There's web work. So we have a mobile work. We have a lot of work there. Then it's about knowing the people. What do they want to do? What is their career path, right? Because the fourth pillar is I have to give you growth, right? And that growth has to be visible to you. You have to know where you sit. You have to know where your next steps are. And you have to feel enough agency in how you get there, right? It's not enough to just say, in my opinion, Here's where you are.

7:10This is the next level. Please go do it, right? Like it's a balance of there are things that the individual needs to do, but then they also need support from the organizational side to get there. And so what we find is if I can give you all those things, if you can see your future here, if you feel that you have control over that next step in your career, right, whether a promotion to senior engineer or engineering manager or staff engineer, whatever it is, right? If you're being paid fairly, if you love what you do, if your work environment is pleasant, right? Those are kind of the structures that we put in place to help with that retention and that scaling.

7:47Given the seniority of the team that you're focusing on and the strategy you're describing, it sounds like you gave your people a lot of autonomy. 100%. I mean, we believe in full autonomy for our teams, right? They know what needs to be done. Now, there's a challenge there, right? At the end of the day, whether it's from a process perspective, what do they run? Do they run Scrum? Do they run Kanban? What's a program increment size? How do they deal with things? They know the answers. This is where we structure a lot on, are we delivering value or not? If we're delivering value, then it's up to those teams to decide the most efficient ways to work to do that effectively.

8:24Now, there are a lot of challenges around that. If you look at introductions of technologies or frameworks, building things different ways. When do we build into a platform-type model and when do we take on tech debt to build a silo, right? Where we've struggled as we've grown to proactively tackle some of those. And so there's been some things we've had to go back and kind of redo. But at the heart of it is autonomy. It really allows us, in an alliance with Chingenta's business model, right? If you look at agriculture, agriculture in Brazil is very different than the US, which is different than Europe, which by the way is different in almost every country, different from Ukraine, different from smallholders in India and Vietnam and the rest of Asia Pacific, right?

9:09So all of these products as they're looking at their markets need to be empowered to take the decisions that are right for them in the moment to kind of seize value. So how do you figure out if you're doing that efficiently and making your team work to the best of their capacity? For us, this is where the metrics that we've introduced are helping, right? I mean, we focus on three things in the metrics. Are we working efficiently? Are we working effectively? And are we working safely? So efficiently is all about for the people and dollars we're deploying, for the effort that we're deploying. Are we using it in a way that is valuable?

9:46Are we not sitting around waiting for things? But all that efficiency doesn't matter if we're not capturing value in the market effectively. And so that's where we expand on things like the door metrics with other things like usage and KPI metrics from products in Amplitude, an analytics platform to understand, are we having the impact in the market? And then the last one is safety. Our industry, like many others, deals with a lot of user data, a lot of grower data. And so as you look at privacy, as you look at regulatory, as you look at those sorts of things, we have to make sure that you can be fast and you can deliver value.

10:22But if you're doing it in a way that is unsafe or not compliant with the laws we need to be compliant with, none of it matters. And so for us, these metrics have really given us those dimensions across it. What kind of results have you seen? It's been great. We've seen things like an 80 % reduction in cycle time this year. Wow. We've seen a 33 % increase in planning accuracy. And we've done this not from a top-down mandate. A big focus on all of our metrics was, as we looked at the basket of metrics we were going to bring in, what actually supports those things? What supports being agile? Not doing agile, right?

11:01What supports steady, sustainable feature delivery? What's healthy for our teams? And paints that picture of are we effective, efficient, and safe. And then we gave these tools to our teams. And say, hey, look at this. right? If there's something you're unhappy with, please move the needle on it. And they were able to. And it's forced a lot of really good discussions on our side, which has been great. Some of it procedural, process-wise. How do we work across the business? How do we work with stakeholders with product, with design to make sure that work is in a better state of ready? How do we make sure that we're reserving the right amounts of capacity for steady state, for run and maintain, so that we can be more predictable in what are we going to do?

11:43When we make a commitment, we can achieve that commitment effectively. And some of those things have led us to better adoption of different tooling, for example. As we look to decrease cycle time, like many enterprises and things, you can't go to production that often. You can't go every day. But it's sped our adoption of feature flagging to allow us from an engineering standpoint to push out and then be able to do a controlled rollout to our different user groups as we do that promotion through production effectively, right? So it's been good on both sides as we rolled this out. So what I'm hearing is you're not just leveraging the analytics side of LinearBee's platform, but you're also using those analytics and the context you get with your teams through these cultural efforts you're doing to say, okay, let's identify friction points and apply workflow automation tooling and other goal setting through GetStream or through the LinearBee platform to actually destroy these friction points and automate them away.

12:38Is that correct? Yeah, I mean, 100%, right? no two teams have the exact same problems, effectively. And so for us, it's really about how do we paint the picture for the teams? How do we help them understand where is there smoke? Where might there be things that we want to go look at and reevaluate? So when we first rolled this out, everyone has an answer for why the metrics are what they are. And we push just a little bit. Let's go reevaluate some of these things. Some of these we can't. So for example, simple things like cycle time's long, QA's on a different continent, this introduces a big delay.

13:16Okay, let's change it. And this gets down into the autonomy of the teams. We're pairing this metrics work with a lot of organizational change on our side. So we have a big focus on team topologies framework. Fantastic framework. If you're familiar with it. And so we're really pushing a lot on just because we've done something in the past doesn't mean we have to do it that way going forward. And the teams are 100 % in charge of that and free to make those changes, right? So, for example, one of the metrics, some of the things we've started to expose to teams is investment level, finance. Oh, wow.

13:53So, you're showing investment metrics for like what products are being worked on, which is like a really cool part. Product level start, right? So, for example, you would have teams that, and I get this as an engineer for a long time, right? Yeah. They always want to try to solve the problem with what they have at hand. This is always what they're looking for, right? And by the time that we get up to leadership of, we actually need two more people for this for 90 days, whatever it is, right? It's too late. Why didn't you tell me this earlier, right? And part of it is they didn't realize and have that product level investment visibility down to them to understand what are our options, right?

14:33And so one of the things we're starting to expose along with these metrics is, here's your budget for the year. Here's your current burn rate. Here's what you have. And so you can take some of those decisions to say, hey, I want to bring on some additional resources for six months. Or I want to move QA from across the globe to a nearer location. What are the cost implications of that? Work with our product and business teams to understand, hey, is this a move we want to make? right? This is so fascinating for me because most teams I talk to that are using linear B's investment profile metrics are leveraging at the board level or with their C-suite and they're saying, oh yeah, like here's what we're doing, you know, let us know your feedback or like here's how we're delivering it.

15:16But you've taken this other approach because you've built these senior autonomous teams and the leadership within that, you're saying, hey, we're going to give you the input. Like, let's make sure you feel this passion. And I think that really goes back to something we talked about a bit earlier, which is the need to make people feel like their work matters and like they can make an impact. And you're giving them all the tools and information to do that, which is a really cool bottoms-up agile approach. 100%. And we track everything across these projects very granularly. All of the people allocations, all of the cloud costs, license allocations, everything.

15:47Because as we look at an investment profile for a product, that product is free to spend their money to achieve their objectives. And so we're trying to put more of that data and more tools in the hands of the teams so that they can be, as we look at team topologies, they can be more autonomous in those changes. They know what changes they can make. And it's not only later when the pressure's really on are they making those changes, they're starting to see some of those signals earlier of where might things not be ideal. And then they're able to proactively make those changes. That's fantastic.

16:22Where are you seeing teams leverage some of those tooling like GitStream's programmable workflows, for example? Our usage of some of the automation stuff like Worker being GitStream have been really valuable from a cycle time standpoint for our teams. So it's really interesting if you look into the cycle time, like the depth of cycle time, like code and the breakdown. All of the metrics, all of the thresholds, are things that all teams would agree on. No team would say we want our PRs to sit for three days and nobody looks at them. And no one would say we want our reviews to take two days. but often they don't know and especially as we look at current working climate one, we're a globally distributed company but also everybody's really for the most part remote first now even if you're in a hybrid environment you still spend a lot of time remote and when someone puts a PR in does the team know?

17:20even if they ping somebody in Slack you get the dreaded, yes, someone will look at it And so for us, where work of being GitStream has been really valuable there is, one, just providing a prompt to the team so they know, hey, there's a thing that needs to be done. But also the reminder when it's approaching the threshold that they're setting effectively. And then for GitStream, the other big part of PR is you don't know if it's going to take three minutes or 30 minutes. You have people, they try for this flow state. They want to be productive. They don't pick it up when it comes. I'll pick it up later, effectively, right?

17:58So just the annotation of who should it go to, how much time is it going to take, how much risk is in this. Our teams have found a lot of value in using those together to kind of just keep on top of some of those things around that review process. It's fantastic to hear your teams getting so much value out of that because it's the thing that we imagined when we started this that we could help provide. So wonderful to see that. I'm also really curious. So I love that you're providing these investment benchmarks to the team and investment metrics, because I think that your approach is fascinating and really useful, where you're creating self-organized teams that have goals and are excited about achieving them and have autonomy and drive and seniority to do it.

18:34What are the other dashboards? What are the other analytics pieces that your teams are leveraging the most, do you think? Outside of? Outside of the investment profile piece. In Linear B? Yeah. We tried to not have them in there. Interesting, okay. Yeah. So this is where you're leveraging your other dashboarding. Our Power BI dashboards, yeah. Yeah, and the reason that we find is that it paints a piece of the picture. So when they want to focus on the delivery piece, those dashboards are super valuable there. So for example, when they want to dig into why is planning accuracy low, and not the subjective part of why planning accuracy is low, because everyone has their opinions.

19:16When they want to understand, okay, what didn't we get done? What came in that wasn't there? It's a great tool for that. But when we start to evaluate, are we having the product impact? Okay, now that work is all in amplitude for us effectively, right? Are we doing it safely and securely? Okay, for us, this is all in Sonar Cloud. Are we meeting our fin obstacles against cloud costs and things like this? This is another tool that we run, right? And so for us, we aggregate all that together into high-level dashboards with Power BI, utilizing the APIs that all these platforms have, right? I mean, this speaks to the configurability piece and the need for that extensibility and how important it is whenever you're picking something for your SDLC.

19:56Huge. And it's been really, really valuable for us, right? Because the other thing it lets us do is create really diverse aggregations of the data, effectively, right? Interesting. And so what it lets us do in this really distributed way is at the bottom level, teams can get the picture of their metrics, right? At the pure product level, or for some of our larger projects, the squad level within there effectively, right? Regions can care about only products that sit with them. We can roll up to global and we can always look at the right aggregation level of metrics to understand, are we doing well?

20:32Are we trending in the right direction? And then also the big thing for us from a BI perspective is outlier detection, right? Is the aggregate metrics may look great, but that doesn't mean there aren't teams that should be paying attention to different things effectively, right? And so the API access and the ability to get data out and marry it against other systems' data from our side has been hugely, hugely valuable. I love the philosophy you're taking here, too, because you care so much about the culture, and it's such an important part of your strategy as a leader, saying, we want senior teams.

21:05We want to build up people up, retain them. And you've very clearly been intentional, like, we want teams. We're not going to measure you on individual metrics. We're not going to screw you over by focusing on velocity, which just gets people to game it. We're going to say, how can I enable my team to deliver and feel passionate and build a unit that delivers for Syngenta? 100%. The dangers of metrics are huge. Everyone's been in an organization that, whether it's intentionally or not, weaponizes metrics against teams and people. Totally. End of story. And this is so much of our focus is on the teams.

21:40Is on how can the teams use these as tools to get better. It's why at a higher level, we look and care about aggregates and trends. Because the risk is if you put any doubt into their mind that these metrics will be used for evil, you lose them. And this gets into, I talked a bit about it today, ownership versus compliance. If you get them to own it, if you get them to care about why are we looking at these metrics? How do these metrics align to the way they want to work? And how are they a benefit to them to care and look at these metrics effectively? you'll see movement. If you don't, you just get compliance.

22:18Velocity is a great example that you brought up, right? Of everywhere. People just game velocity. It's a dangerous metric. They just make the points work. We've all heard it. I'll bring in, oh, this one's a soft two points. So bring that in to make sure that we can get the number that we need. And I would just make that one a little bigger. And, you know, it's not a valuable tool for anyone that effectively, right? And that's the risk to me that any introduction of metrics brings in and why how you use it and how you talk about it is so important. And honestly, I think this is why we at Linear B see Syngenta as a really fantastic partner.

22:53We share this philosophical belief in the value of engineering teams and the avoidance of this over-measurement of individuals saying, you know, you presented earlier today with Ori Kiran, our CEO, a fantastic presentation on Syngenta's transformational push and how you're aligning engineering metrics and workflows and cutting down on blockers, which I want to get into in a minute here. but my biggest takeaway was how philosophically aligned we are around this perspective on how engineering orgs need to do this. Yes, metrics are dangerous. It's also, frankly, it's a necessity today. You're not going to get away with it if you're an engineering leader.

23:24We have to be the ones to define the conversation. Otherwise, we risk the McKinsey's of the world coming in and saying, focus on these individual metrics. It's a big reason why when I talk to Ori, he's like, yeah, we can talk about developer product to you. That's the phrase a lot of people use, but let's talk engineering efficiency. How do I create high-performing teams that are more efficient? And developer productivity is nonsense from a metric standpoint. And no matter the metric you choose. And it's as simple as, you can have people working on incredibly complex subsystems in a team, and another person who's churning out controllers and APIs and things like this.

24:00By any objective measure of changes of code, of anything like this, one person looks 10, 15 times more productive than the other. But it doesn't mean that that's what's happening. until we come up with an objective measure of value delivered somehow, there's no real way. And that's why we focus so heavily on teams, effectively. If anything, the value I find in the developer-level metrics is where do we need more support. I think Martin Fowler wrote an article 15 years ago where he's like, you can't measure developer productivity. And I really see that today. People are still fighting that. They're like, well, I don't know if I take it as far as I was like, no, like help your team succeed, build high performing teams.

24:43And I think it's my really admire about the approach Syngenta has taken. And I know as you've scaled, you know, as we mentioned at the start of the conversation, the last over the last two years or so, you've built your team from 150 to 400 developers. I know you've ran into a lot of blockers in that time frame. And especially as you kind of are in that hyper growth, everything starts to become a blocker. What were the problems you're running into and how did you address them? I know we've talked about some of them already. So for us, it's the evolution of small to big, especially as our products got bigger and things like that.