Lenny's Podcast: Product | Career | Growth - An inside look at Mixpanel’s product journey | Vijay Iyengar (Head of Product)

Starting point is 00:00:00 The issue for us at the time was that we took people away from the investment in our core product to go do those other things. We moved people, right? And so the trap there is that you leave yourself right for disruption in your core because someone else can out-invest you in that core. And so if you're the leader in some core product, our takeaway here is you should continue to out-invest everyone else in that core and then invest the profits that come out-of-that-cor into the next venture. Like invest profits and not people or venture capital. which is maybe like net present value of profit or something to that effect. But don't take people away from the core to go to those other things because then you end up distracted. Welcome to Lenny's podcast where I interview world-class product leaders and growth experts to help you get better at the craft of building and growing products.

Starting point is 00:00:47 Today, my guest is Vigéy, Iyngar. Vaj is currently head of product at Mix panel. He actually has a very similar career trajectory to myself, where he started as an intern at Amazon, then he was an engineer for a while at Uber. then he became an engine manager at Mix Panel, but then he shifted from an edge manager to director of product and now head of product at Mix Panel. You don't often see people moving from an engine leadership role straight to director of product. So it was really interesting to hear what he took from his Eng experience and brought into his approach to product leadership. But we

Starting point is 00:01:17 spend the bulk of our time talking about what he's learned from the journey that Mix Panel has been on, where they started with a simple product, then scaled to a number of different products, solving many problems for customers and then made the hard decision to scale back to just a single core focused analytics product. We talk about why they made that choice, what they learned about when it makes sense to expand to new products and when you probably shouldn't, and how they approach to that organizationally. We also talk about how mixed panel built product, how they think about product philosophy, how they prioritize, and also what you're probably doing wrong in how you set up your analytics for your own product. With that, I bring you Vajaiy, Ingar, after a short

Starting point is 00:01:52 word from our wonderful sponsors. This episode is brought to you by Pan Pando, the always-on employee performance platform. How much do you love the performance review process? Mm, yeah. It's time-consuming, subjective, biased, and there's rarely any transparency. With the rapid shift to distributed work, it's a struggle to create the structure and transparency that you want to help your employees have the highest impact and growth in their careers. Hando is disrupting the old paradigm of performance management,

Starting point is 00:02:23 including a continuous employee-centric approach, so employees stay engaged, see their progression in real time and know exactly when and how they can level up. With Pando, managers can leverage competency-based frameworks to effectively coach and develop their teams and align on consistent growth standards, resulting in higher quality feedback and higher performing teams. Visit Pando.com slash Lenny for more info and get a special discount when you sign up and reference this podcast. That's pando.com slash Lenny.

Starting point is 00:02:52 This episode is brought to you by Notion. If you haven't heard of Notion, where have you been? I use Notion to coordinate this very podcast, including my content calendar, my sponsors, and prepping guests for launch of each episode. Notion is an all-in-one team collaboration tool that combines note-taking, document sharing, wikis, project management, and much more into one space that's simple, powerful, and beautifully designed. And not only does it allow you to be more efficient in your work life, but you can easily transition to using it in your personal life, which is another feature that truly sets Noion apart.

Starting point is 00:03:28 The other day, I started a home project and immediately opened up Notion to help me organize it all. Learn more and get started for free at Notion.com slash Lenny's Pod. Take the first step towards an organized happy team today. Again, at notion.com slash Lenny's Pod. Mijay, welcome to the podcast. Thank Lenny. Great to be here. A huge fan of the pod. So glad I can contribute.

Starting point is 00:03:57 I definitely want to talk about Mix Panels journey, both as a product. team and a product. But before we get there, as an engineer, you're in a long-time engineer and then you became product leader. Is there anything you had to unlearn as an engineer and the way you thought about leadership and product and business? One of the things, after you've been in engineering for a while, is that you've developed this tendency to immediately respond with no to new ideas. And I think in the engineering perspective, this is because you spend a lot of your time building and maintaining ideas that maybe are half thought out or didn't really go anywhere. And you just feel like the full

Starting point is 00:04:31 brunt of the maintenance cost of that. And so you build up this scar tissue and this immune response, the same no to new ideas. And it's a hard no. Like, no, we're definitely not going to do it. And I think I had to unlearn that moving into product because, you know, you get a lot of ideas coming from a lot more places in the organization. And ideas are fragile in their infancy. And it's, you know, a hard no can really kill a whole direction that you could potentially

Starting point is 00:04:55 go. They could be very high reach and high impact. So one thing that I found is the best way to get to a no. if you ultimately need to get there is to try to make it work, like start, try to make yes work and document how you've tried to make yes work and do that earnestly, not as a, you know, as an exercise of just an alternative that you're considering. Try to do it sincerely and get to know after trying to make yes work. And so that's, that's one thing I've been trying to just apply my engineer problem solving brain to do that instead of thinking about how it might not work and something saying

Starting point is 00:05:24 no. Is that something that you recommend engineers work on like looking back? I know as at PM, it's like, oh, I love when engineers say yes. This is awesome. I'm going to help everyone learn to say yes. But as an engineer, obviously, that's a challenge often. What do you recommend to folks that are engineers currently that maybe want to improve on this or shift in how they think about saying yes, saying no when they're asked about something new? I think some of the best engineers that I've worked with actually already do this.

Starting point is 00:05:51 But Fultor, they're able to balance both in their head. So ultimately, it's this balancing act. You just want on call. You were woken up three times at 3 a.m. due to various bad ideas. and the next morning you wake up and stand up, it's like, hey, can we do this new thing? You're, you kind of have to have that empathy and do that. So yeah, I think the exercise is just like, just take 10 minutes to consider the idea. And just sincerely consider how might we make it work.

Starting point is 00:06:13 And if at the end of those 10 minutes, it's like futile and there's no path, it's fine to say no. And it's a good, good instinct to say no, actually in a lot of cases. But yeah, I would recommend that to engineers. I think it would have been better in my career for sure if I'd learned that sooner. I want to spend some time on the mixed panel product journey. It's been an interesting roller coaster. I think the company's been around for how many years since 2010, 2009? 2009, yeah.

Starting point is 00:06:35 So I know it started as kind of a very simple product analytics product back in the day. And then as you do with ambitious companies, you look for more problems to solve. You look for more problems to solve from your customers. So as I understand it, you guys added a lot more products to the suite of mixed panel products. And then I know that there are some challenges with scaling that and maybe the products didn't stick as much as you were hoping. and what I understand is recently moved back to just a single core analytics product. And so I'd love to just hear that journey of what that process was like, which you learned as a product leader and as a company,

Starting point is 00:07:08 about kind of scaling, expanding, trying to solve a lot of problems, and then coming back to one core straightforward problem. McFell started in 2009 as provided product analytics to EPD teams. I think early on it saw a lot of success because it built this in-house database called ARB, which stands for arbitrary segmentation. And that was necessary because events data, which is the fuel for product analytics, is a few orders of magnitude larger than most other types of data

Starting point is 00:07:36 that people collect. And so you need a specialized approach to deal with it. And so that, I think, spurred the first wave of explosive growth because product analytics was a really burning problem at the time. People were shipping mobile apps like crazy, and they needed a solution that could scale, and that was kind of a durable mode for McFanel for a while. And I think because we had this SDK that was installed in so many apps,

Starting point is 00:07:55 and we have this really scalable event collection and analytic interface, it was just natural to expand into a few adjacencies that would leverage those same technologies. The first one was messaging, being able to send targeted messages to users, which is something that's fairly naturally you might want to do,

Starting point is 00:08:11 especially if you have an SDK already installed. Yeah. The other aspects that we've expanded into was data infrastructure and trying to be sort of the single source of truth of data in companies. And what ended up happening was that by 2018, we had this big churn problem.

Starting point is 00:08:28 We had something like 40% churn, revenue churn, our core product. And when we dug into it, it wasn't that people were shurning because they didn't need product analytics anymore. They had the need. They were just churning to competition because we were just not up to the market

Starting point is 00:08:40 in terms of the features we had in our core. And when we dug into why that was, it was just that we had a 50% engineering team that was building products across three domains, product analytics messaging and data infrastructure stuff. Our engineering team was just spread too thin to address all of those core gaps in functionality. And so we made a really hard call at the time.

Starting point is 00:08:59 We said the hard no to those two other categories and decided to focus our entire engineering team on closing the gap on product analytics and innovating there. And from a process standpoint, how we operationalized this was we threw away all our planning, all the execution and the work that we'd plan to do so far.

Starting point is 00:09:17 And we did something very simple. We took all the churn reasons that our customer success and sales teams had been painstakingly collecting for years, grouped them by category, which was like roughly product features we needed to build, sorted descending by ARR, took the top 10 things and made that our roadmap.

Starting point is 00:09:32 I just gave every engineer, you know, direct access to customers and gave them a bucket to go work on, which I think goes against about a million product best practices out there of just doing that. But I think given the context at the time, we need to optimize for speed. And speed comes when you have extreme clarity on what you want to do and focus.

Starting point is 00:09:50 And so we really just, optimized for speed in that time. And so in that first year, we moved really quickly and we shipped something like 100 features in that year and closed a lot of gaps. Again, these are all vanity metrics, like measuring number of features doesn't mean anything. And what year is this,

Starting point is 00:10:05 by the way? This is 2018 to 2019. Yeah, so we moved really fast shipping all these features and instantly saw the improvements to win rate and retention. But one of the cracks that started to emerge was we neglected the holistic design of our product at the time, right? And if you're, if you're shipping features that quickly, it's, you don't have time to stop and think, like, where does this go and how does this fit into

Starting point is 00:10:27 our overall system architecture? And what started to happen was that we were hitting diminishing returns with some of these features. And, like, not considering the holistic design and consistency meant the reach of every feature was low, right? Like, you had to rebuild it for every part of the product that we were in. So at the time we made, we spent up the second stream that was very design-led. And I think this was also coincided around the time we adopted Figma. And so really brought design at the seat at the table of the company. And we just set up this goal to make design one of our key differentiators. So this design-driven initiative was really about how can we think about the system architecture of our product? What are the key building blocks of McFanel? Where do they

Starting point is 00:11:01 need to fit? How few of them can we have, which is a really important step, and then how old users discover them and how do they relate to each other? But I think this realization was born out of the fact that so many great products would are lose based on their architecture. I think Notion, for example, that pages and blocks architecture is so strong and you can hang so many features off of those core building blocks in a way that has such high impact on a region and a discovery of those features. So anyway, we did that in parallel with the and continued that grind on core gaps. And so the end result of that phase, which was from, you know, 2018 to maybe late 2021, 2021, 2022 was our retention went from about 60% to 90% and our NPS went from 16 to 50.

Starting point is 00:11:43 So I think, yeah, I mean, there's a lot in there to unpack, but repocusing on the core really helped us achieve those results. Got it, yeah. I have a lot of questions about this. So, interesting. So that phase that you went through where you sorted things by potential ARR, was that the phase of expanding

Starting point is 00:12:00 to multiple products, or that was post, we're going to focus on analytics and go all in there? Oh, that was post-focusing on analytics. Okay. Yeah. And you're saying that you had a stream of just build all the features

Starting point is 00:12:12 that were lacking, that are causing customers to churn. And in parallel, there was a track of, let's build this product such that it all connects and works together well, and it's really well thought through long term. The first thing I might have made it seem like it was just the buckets for features.

Starting point is 00:12:27 We did take the step of turning them into problems and being clear, exposing engineers directly to the customers that had those problems, and then invent a solution to solve them. So, I mean, loosely, there were features involved, but a lot of them were kind of core problems we needed to solve. But first approach is so interesting. Like, it's kind of like, yes, we will make more money if we focus on these features. To your point, it ends up being just a bunch of features and products that kind of maybe

Starting point is 00:12:49 don't synergize. Looking back, was that a good idea to approach it that way, at least for a while? It highly depends on your context. In a very competitive context where there are just table stakes features that customers need and that it's been validated by the market, you need to optimize for speed more so than anything else. But it is an approach that outlives its usefulness pretty fast. And we put that approach behind us relatively quickly after that phase. And I would actually say that design-driven phase was the next phase, where it was, okay, we're not bleeding on the table stakes anymore, but we want to make a holistic product that has high reach, high impact on the features and is actually usable. And so that was, I think, a follow-on phase that's necessary. Obviously, depending on your particular circumstances and competitive dynamics, you can sequence them differently.

Starting point is 00:13:35 But I think it was the right call to just sort of like that on-call thing again, where, you know, when you're in trouble, you've got to get out of trouble. you can't mow your lawn while your house is on fire. You kind of put out the fire and then deal with everything else. So that's kind of the approach we took. What's an example of feature or product that you launched within that first track? And then what's an example of something that came out of the designer-led approach, if anything comes to mind? I think out of the first track, oh, man, there's so many that were just core.

Starting point is 00:14:03 Like, we didn't have a good cohort product at the time. Like, just being able to create behavioral cohorts of users and create them from any report that we built, right? I think that, I mean, it's just table sticks and analytics to be able to do that. So that was one of the first things we built and it was fairly obvious. There's a lot of other things in like more bad steps of funnel analytics and a flows, a visualization that was, you know, really interactive. I think on the design-led phase, the biggest thing I think was visualization consistency

Starting point is 00:14:31 and making our charts interactive in a consistent way across all, across our entire product. And so there's two things that enabled. One was that every time we added a new visualization or a new enhancement to a visualization or like how something is sorted in one report, it just instantly applied everywhere. So just the reach was multiplied for everything you added. And the other thing is it just made the product more accessible. Let us add dark mode. So it made our visualizations really stunning and really easy to see what the takeaways were. And then every new visualization we added inherited all those benefits. I'm trying to think about like being at a company that goes through this phase

Starting point is 00:15:03 of, hey, we're just going to build a bunch of stuff that we know we need. And it feels like hearing it. It's like, oh, yeah. And then we're just going to make it all look great and connect. and work well. I imagine that wasn't planned, and I imagine that wasn't easy to get people to maybe slow down on just building more products and features or push it into direction where it's all going to make sense. Can you talk at all about what that process was like, like how hard it was to shift from, we're just going to knock through all this checklist of things to like, let's just figure, let's slow down, let's spend a lot of time designing. It was definitely more messy internally that I described it.

Starting point is 00:15:38 One of the key junctures was when we had this really talented design team, and we were putting them on these very tactical projects that was, frankly, that was very engineering-led, right? And design would often come in at the end and be asked, like, hey, can you just make this look nice and put some pixels on it? And it's just such a waste of your design team

Starting point is 00:15:55 to have them do that. But at the same time, the pace was so high that they didn't have time to come up for error and do anything else. And so there's actually this moment where I was an engineering manager part of this, and we had a meeting with our PM and our head of design at the time, and you said, hey, we can actually do the next three months of projects about any design,

Starting point is 00:16:14 which was a kind of controversial thing to say. But we're doing this so that you can take three months with a set of designers and go think about the system architecture of the product, and we'll wait for that to be done before we do any architectural things that might impact the architecture. And I think that gave designers a bit of breathing room to go do that, just separating them for a bit from the tactical fire. because what was happening instead was

Starting point is 00:16:36 we would get towards the end of the project, bring design in, and they would use each project as an opportunity to squeeze in like, oh, and we can simplify here, and that's just a classic way to blow up scope at the end of the project, because there wasn't a dedicated space for design-led projects. And I think that was kind of a key friction point

Starting point is 00:16:51 that we ultimately had to decouple for a bit, and then regroup and say, okay, now what's our strategy? And just take on projects purely for the sake of improving consistency, reach, depth of our UX. Also looking back, the process you went through adding a bunch of products to solve more customer problems, something every founder and product team thinks about.

Starting point is 00:17:12 When should we add new product lines? When should we expand beyond the core? I'm curious what you take away as a lesson and what you advise other founders and companies when it comes to when is a time to expand and think about a new product and a third product and a fourth product. I don't know if there's a hard and fast rule here. I can just maybe say what made sense

Starting point is 00:17:31 and didn't make sense in our context. The issue for us at the time was that we took people away from the investment in our core product to go do those other things. We moved people, right? And so the trap there is that you leave yourself ripe for disruption in your core because someone else can out-invest you in that core. And so if you're the leader in some core product, our takeaway here is you should continue to out-invest everyone else in that core and then invest the profits that come out-of-that-cour into the next venture. like invest profits and not people or or venture capital, which is maybe like net present value of profits or something to that effect. But don't take people away from the core to go to these other things

Starting point is 00:18:08 because then you end up distracted. And the other thing aspect of that is that those secondary products we took on were in categories of their own. And it's really tempting and you'll often get dragged into building, you know, accidentally entering another category. And then you'll end up building these Bolton products that are the end best in their category. Right. And like the adjacent categories for analytics or like CDPs or,

Starting point is 00:18:29 message targeting or feature flagging or something. But there's not that many people that need the sixth best CDP or the eighth best feature flagging or the 10th best message targeting tool. And it ends up being, you know, in aggregate will contribute 5 to 10 percent to your revenue, will seriously accelerate your growth rate and then takes engineers away from the core product. And so those were the circumstances that we were in. And I think if you're seeing churn to your competitor on your core product and you're not best in class on any of the other ones, then maybe it's time to reevaluate.

Starting point is 00:18:58 And then the last thing I'll say there is that it's also 10x more painful than you think to cut mild successes than anything else. And just organizationally it's painful. And there's teams that have a whole roadmaps. And it's a really painful experience. So you have to think really hard before you kick those off. That is really, really insightful advice. Makes me think about if you bundle good enough solutions, there needs to be kind of this anchored tenant that like I will not give this thing up. And I'll use like the third best version of something else if you have it.

Starting point is 00:19:30 But if you're not that valuable and important, you're not going to convince people to use something because you're competing against the best in every category. Exactly. Yeah. That is really interesting. I've been doing this kind of series on how different companies approach building product. And I have a few questions I'd love to ask around the product development process and Mix panel. Sure. The first is just like, how do you plan?

Starting point is 00:19:53 How do you plan? I know it evolves, but just how do you plan currently? Like how long are your planning cycles? How far ahead do you plan in detail? Do you use OKRs? Maybe I'll start there, those three kind of sub questions. We have these kind of unsolved problems in analytics that we're going after. For us, that's like, people always want more power, more simplicity, better data trust,

Starting point is 00:20:13 faster onboarding, better collaboration, better price performance. And so we largely organize our teams around those problems and those missions. One quick aside there is that, you know, some of those problems have attention with each other. like power and simplicity, there's a trade-off there, right? And we want one team to own both so that they can, they're kind of forced to confront that tension and beat that trade-off. And so that's kind of how we think about generally our product team is these cross-functional EPD teams,

Starting point is 00:20:42 each of which that's focused on solving these long-lips-paired problems. Procaths or like our core analysis team focused on that power simplicity trade-off problem. In terms of planning, the way it works is that we plan on a six-month time horizon. And I can talk about our most recent planning cycle, actually, because we're just completing it. Yeah, let's do it. Yeah, basically, it started out with this strategy memo that our leadership team wrote

Starting point is 00:21:06 that basically just conveys this is where we want to go as a company in the next year, and here's that the product team can contribute both of that. I just established these key pillars. We shared that with the teams, and they took that and also combined that with all the quantitative and qualitative context. They're constantly consuming about the problem they're working on and our customers and ideated and developed the series of bets for the next six months, which I think are, to some extent, similar to OKRs,

Starting point is 00:21:29 where the anatomy of a bet is that it's problem we want to solve, our hypothesis on the solution, and then some plan to win, like some plan to actually get there and a way to measure that you got there. And take one of the unique things that we did relative to other companies that do planning is I think it usually is sort of this W process of there's the strategy memo, and then teams generate bets,

Starting point is 00:21:51 and then there's a review, and then they go back and iterate and then they finalize. And we kind of collapsed the middle part of the W where myself and our head of design actually spend time with each of the teams actually ideating on the vets and participating in the solution discovery process going into the jam sessions

Starting point is 00:22:06 and adding fig-masticies ourselves with ideas and dots on things, which we did because we aren't a huge product team and we're not going to do like 50 things and a half. We're going to do maybe 10 to 12 things. And so that's enough that we don't, We can do something that doesn't scale if that enables high bad with communication between

Starting point is 00:22:24 us of the teams. And it ends up being more messy and unstructured a bit in that phase because we're just, we're in their contributing ideas as well. But by the end of it, I think the team leaves feeling both more competent in their bets because there's more thought that's gone into it and then more aligned up to bottom one, you know, why we made certain decisions. And so I think that that's one thing that's different. And then we conclude that process with the road show where we presented the rest of the company

Starting point is 00:22:46 and get their feedback as well. How long is this process fairly? The teams did pre-work for a couple of weeks, like two weeks in December, and then we did a two-week sprint on solutioning and ideation in January, like the first two weeks of January. Awesome. And what's the end result of planning for each team? Did they deliver a document with like, here's our strategy, here's the big bets, here's a roadmap? Is there a template you pass around? How do you kind of get to a thing that people share and present and comment on? Yeah, I think there's basically three artifacts that are kind of linked to each. other. So the first is we use Notion. And so we have a database of Notion called Betts, which is where each page in the database is a bet. And that has a template. Yeah. So it's like kind of roughly what I described with a few more sections of what problem we're resolving. What's the

Starting point is 00:23:35 evidence of demand? What's the reach and impact of this problem? How do we know are successful? And then what's the key driving hypothesis behind the solution? And then a rough plan. And then that's tied with the presentation that's kind of like a tight summary of that that has like one slide per bet. and then is also tied with more of an execution focused, how do we sequence and staff this thing and eliminate dependencies, which the engineering team contributes to. So I think those are the three artifacts that are linked together.

Starting point is 00:24:01 This episode is brought to you by lemon.io. You've achieved product market fit. You're able to activate, engage, and retain your customers, but you don't have the engineers that you need to move as fast as you want to because it's hard to find great engineers quickly, especially if you're trying to protect your burden rate. Meet lemon.

Starting point is 00:24:20 Lemon.io will quickly match you with skilled senior developers who are all vetted, results oriented, and ready to help you grow, and all that at competitive rates. Startups choose lemon.io because they offer only handpicked developers with three or more years of experience in strong, proven portfolios. Only 1% of Canada to apply get in, so you can be sure that they offer you only high quality talent. And if something ever goes wrong, Lemon.io offers you a swift replacement so that you're kind of hiring with a warranty. Learn more. Just go to lemon.io slash lenny and find your perfect developer or tech team in 48 hours or less. And if you start the process now, you can claim a special discount exclusively for Lenny's podcast listeners, 15% off the first four weeks of working

Starting point is 00:25:07 with your new software developer. Grow faster with an extra pair of hands. Visit lemon.com I know you have some insights on prioritization and some strong opinions on how to prioritize. Can you talk a bit about that and how you advise your product teams to prioritize? One really common framework in prioritization is Rice, Reach, Impact, Confidence effort. And I think it's simple and fairly robust, which is, I think, generally good qualities of a framework. But one of the traps with Rice that we observed is that the C and E, the confidence and effort, tends to cause you to prematurely deprioritize potentially high-reach, high-impact bets,

Starting point is 00:25:46 really innovative things. And we encountered this on one of our teams early last year where we just riced everything, all the ideas, and a lot of the high-reach, high-impact things ended up at the bottom because confidence and effort were just so murky for them, as they should be typically for high-reach, high-impact ideas.

Starting point is 00:26:02 And to one exercise that we push our teams on is just ignore the C&E for a little longer than it's comfortable, and just sit with those high-reach high impact ideas with like engineers and designers in the room committed to actually trying to solve them, like give it a fair shot. And you'll often find like if you spend a week on that set of ideas, you can get pretty far in understanding the confidence and the effort. You can probably find a higher, confidence, lower effort way to do them, then add the C&E back in and then

Starting point is 00:26:28 rice as usual. And the goal is to end up with a reasonable mix of innovative bets, incremental bets and then ones that are, you know, technical debt or product debt that you need to address. I usually just cut out the scene myself. I find that it's not that powerful. Do you do this in a Google sheet? Do you use this in Notion? How do you actually recommend teams do this predization or just eyeball it? I'm not super opinionated on the exact tools that teams use. I think this is like a team local exercise typically, but most teams use Notion and just civil tables or databases inaction. Sweet. I think the other thing on partization that's always tricky is estimation. And every engineer will tell you estimates are all lies.

Starting point is 00:27:07 And if you say it'll take eight weeks, it'll take eight months. But one, I think the core problem with estimation is you're asked to estimate things before you know what the thing is. And it's just a strange output to be expected to produce. And one approach that I found really interesting is from this book called Shape Up by Basecamp, which is this idea of appetites over estimates, where instead of making the estimate an output of planning, you make the time box or an appetite, the input. And you say, like, we want to solve X problem and we're willing to invest six weeks,

Starting point is 00:27:37 solving that problem. The obvious question there is like, you know, how do you pick that time window? It just seems arbitrary. And so the base camp people suggest just pick six weeks for everything. And they're really austere about like, if you can't scope hammer or something down to six weeks, you're doing it wrong,

Starting point is 00:27:52 which I think is, it can work and has a lot of benefits that you, like creates a rhythm in your company. But one approach I found that works better is pick a reasonable sounding appetite and just explore the two to three options around it. Pick six weeks and then say, what would we do differently if we only had four weeks?

Starting point is 00:28:07 or eight weeks. And you'll kind of naturally find the efficient frontier of, you know, cost and an impact and then align on that. And the important thing is that you check in after that time period and say, is there any new information that's just we should continue? Did we uncover the biggest risks? And are we just on the long tail of things? And actually be honest with yourself about that. I think that's important regardless of what framework you use. I really like that. So does that how you actually operate? You create a time box. We have four weeks for this project, whatever we get done, we ship, whatever we don't, we push up. We operated that way in engineering, like, literally on the infrastructure side, because we had this

Starting point is 00:28:41 series of projects that would just take forever, and the longer it takes, the longer it's going to take. And so we've done that exercise quite a bit. I'd say more on the more engineering heavy projects than others, but we're starting to adopt it more in the product side as well. The main exercise we've taken on the product side is more the consider what would you do differently with different time boxes approach. Just a thought exercise. Yeah, it's a good thought exercise, and it just forces everyone to truly score the requirements. Like, critical nice to have is nice, but really if, you know, in two weeks you're going

Starting point is 00:29:14 to get pulled off to do something completely different, what would be a complete solution that addresses the core problem? And it forces you to pull like the meat of the problem in first as opposed to just doing the things that are surrounding it. That's cool. I really like that. I've done that myself. I'm curious if anyone ever does the shape up process for real, where it's just like we will

Starting point is 00:29:30 ship anything that is ready within six weeks and not actually have like specific deadlines or kind of like concrete goals of products they need to ship in specific ways. Yeah, well, I think the shape up process, if you run it all the way, they do, their idea is that you can actually predict on a six week time horizon. So you, you can just hammer down scope to something that is complete. It needs to be complete. It can't be milestone one. That's like a half baked thing in six weeks, which I think that rigor, like the rigor they applied to that across the board. You need to do it all the way. You can adopt the process halfway, I think is the challenge. Otherwise, you end up with people shipping like milestone one and then moving on,

Starting point is 00:30:06 which is not the complete product. Makes sense. A couple more questions around how you built product. You mentioned that you have a unique approach to keeping product teams close to customers. And I'm curious what you've learned there, what you found to be helpful and just kind of, yeah, keeping product teams close to your customers. I think this is one thing that is something we invested in pretty early on. It makes that all.

Starting point is 00:30:27 Actually, around that time in 2018, when we, repocused on our core product. One of our sales engineers, Aaron, built this automation where you piped all these customer gaps that we got that were reported by our customer success and sales teams, piped that into Slack and just a feed.

Starting point is 00:30:43 And what this created was this culture where all engineers and designers could consume that raw feed of direct points of customer with no gatekeeper, no process to access it, no pre-aggregation, right? And I think this scale is pretty far. Like, at a product

Starting point is 00:30:59 team of our scale and with our reach of customers, we don't get so much feedback that someone couldn't read it in 20 minutes every day. And for like four or five years in engineering, every day I would read all the gaps that we got. And many engineers would do that. And one of the rituals that it's enabled is we'll find that engineers will go into that channel and react with a message with an email emoji, which means I'm going to email this customer and find out more. And they'll email the customer and say, hey, I'm the engineer that built this feature. I saw you said the specific thing. Can you tell me more?

Starting point is 00:31:29 I'd love to understand. They ask the five wise, and then they improve the product on their own. And I think that culture is just so important. It just empowers all engineers and designers to think like a PM a little bit, which I think takes a little bit of the load off on the PM to be the gatekeeper of all that information. And then over time, we've evolved it quite a bit as our data stack is involved.

Starting point is 00:31:50 So we now not just take customer requests, but we take things that are posted on Twitter and NPS survey feedback. and win-loss notes from our competitive deals and pipe them both into Slack and into Notion so that we can both get the real-time feed and then we can sort and aggregate and tag things accordingly. But the key artifact of this is that it's all open. There's no gatekeeper in that process.

Starting point is 00:32:11 That sounds both amazing and wild. Do you still allow engineers just to email customers and ask them questions about the stuff, or is that harder to do it as you've grown? No, no. We still allow that. Yeah. Wow. That's awesome. One nice thing about the stack, actually,

Starting point is 00:32:24 the data stack is that, So basically all these feeds information land in our data warehouse, which is BigQuery. And from there, they're pushed out via a reverse Ctel tool we use called census to Slack and Notion. Does that make the no code? But one of the benefits of that is that we can enrich all of these feeds with, you know, who's the account? What's their ARR? Who's the CSM? And like all that other contact information.

Starting point is 00:32:45 So it's usually not like an engineer is blindly reaching out to a customer without letting a CSM know if it's like a million dollar account or something. the idea is just like trust them with that context and they can tag the right people and make the right call. I'm so curious how that gets prioritized and how PMs are looped into all that, but we don't have to get too deep in that. That's a really cool process. I haven't seen that before. Our engineers are just emailing customers digging into questions and problems.

Starting point is 00:33:08 The trap, of course, is what you just called out is like you can't be reacting to everything all the time. And certainly if you ship a redesign, right, like the first two weeks of that, there's going to be a bunch of feedback that's like, I hate this, go back. And I think that's sort of an organizational muscle we have to build. to balance the reaction, and that's just a thing we've had to practice doing. But I think the tradeoffs are worth it. Awesome.

Starting point is 00:33:28 One last question along these lines. Can you just talk about the tools you use, like the SaaS products you use to run your product team for collaboration, communication, notes, docs? You mentioned Notion as an example. I think our stock is actually fairly standard these days. So we have Slack, Zoom for communication, Notion for docs, and any long form writing, and it's a wiki and database. and fake magic jam for design and webboarding.

Starting point is 00:33:54 I think what's actually more interesting is our data stack and the productivity we get out of that. I briefly touched on this where basically all our data gets EPL out of all the systems we have, Lanz and BigQuery gets joined and modeled and then pushed out via census to all the other tools in our stack. And I think that's been a huge productivity in lock because you can build internal tools with very little code.

Starting point is 00:34:17 If you can write SQL, you can build an internal tool, basically, and that pushes information to the teams that need it. And so that, I think, just has unlocked a lot of these types of things, like automating qualitative signals with no code in a reliable way. And then if someone is like, oh, can I get ARR on this? Yeah, sure, it takes two seconds to do that. So I think that data stack has been a huge productivity on block for us. Awesome.

Starting point is 00:34:37 If you guys shared that anywhere online, just to show kind of the stack that you guys have built. We have a couple blog posts that talks about our stack for, sort of like, we use this both for kind of our PLG infrastructure and like our product-lit sales. you know, like defining a Ptchul and alerting when you use their fitz that criteria, but then we also use it for internal tools. Yeah, we have a few blog codes on that topic. Sweet. We'll follow up and include some of that in the show notes. Yeah, definitely.

Starting point is 00:35:00 Final line of questioning. You're one of the smartest people in the world on product analytics, heading product for Biggs Panel. I'm curious what you think most people get wrong when they're setting up product analytics for their site, their product, their company. It's maybe a bit of a hot take because I think so much. many people, there we go. So many people still do this. But I think the biggest mistake is setting up analytics using client side SDKs, client side tracking. So like web and mobile SDK is like putting a McFanel. Dot track or segment dot track in your web app or your mobile apps. And the reason is the hot take is that for many people that's product analytics and

Starting point is 00:35:38 SDK tracking are synonymous. They're like, all right, McFanel means SDK. I have to put an SDK in my web and mobile app. But that's a mistake because we've just seen time in time again at leads to poor data quality and difficulty to maintain that data. So the problem on web is just due to ad blockers and other unreliable things in the JavaScript world, you end up dropping 20 to 30 percent of your events. And so it just doesn't match your internal databases. And then on mobile, there's two problems. The first is that you have to reinvent tracking for both iOS and Android, because it's

Starting point is 00:36:07 two different languages and two different platforms, generally speaking. And so you end up with many duplicate events that semantically mean the same thing, but are just different because of the two platforms. and you might have two teams owning that. And the second issue, which is, I think, even worse, is that you are kind of beholden to clients updating their mobile app to get to the latest version that has your latest tracking. So if you want to add new tracking, it'll only apply to people at the latest version

Starting point is 00:36:30 and beyond. Of course, yet, all of your old tracking, whether it's broken or you made a mistake, is still out there in the wild. And so you're constantly getting events that are old and broken. And so what we recommend instead, and though we've seen a lot more customers adopt recently, is just track events from your servers. instead of your clients. And that has three benefits.

Starting point is 00:36:49 One, it's instantly cross-platform, web and mobile and TV and whatever other platform. They're all going to go through your servers, so you instantly get 100% reach. The second is it's in an environment you control. So if you want to update tracking, you can update it. It updates for 100% of your users.

Starting point is 00:37:04 And then the third thing is that, and this is, I think, maybe unintuitive, but it's true, is that engineers have been tracking events from servers forever. It's called logs, right? And events are just logs where they use their ID. in them. And so they don't need to deal with learning in USDK and dealing with all that. They just have to track logs that have some structure and a user ID in them. They're tracking events. And so

Starting point is 00:37:25 if it's easier for the developer, it'll get done in an higher quality way. So I think the really simple advice there is just start tracking events from your servers instead of from your clients. And if you need to supplement it later on with context that's only on the client, you can add that on later, but server side should be the default. Maybe I'll ask question is just what's changed most and how companies work with analytics in the past few years. And then just where do you think things are going in the space of analytics? Yeah. So I think one huge trend is the rise of the data warehouse.

Starting point is 00:37:55 So these are, you know, snowflake, BigQuery, redshift. And they're really scalable. And they speak the SQL standard, which has led to this explosion of tools that have emerged around them and make it really easy and cheap to load data into the data warehouse. And then also easy to push data out at the data warehouse. I think tools like Hansis do that. And this has a few implications. So the first is that the data warehouse becomes the center of gravity for all data in your company.

Starting point is 00:38:21 Whether it's product marketing and sales data, they all land there. And I think that's really valuable today in this product-led growth world. And a lot of ink has been filled about that. But from a data standpoint, that means, you know, all these teams need to be operating off the same version of the truth. And that version of the truth is sitting in the warehouse. And it just needs to be joined correctly. the second thing in terms of where things are going is that events,

Starting point is 00:38:45 like a time series of users did at this action at this time, are the universal data model for analytics. And the reason for that is every action, every interaction a customer has, whether it's with your sales team, like a gong call, or with your marketing team, like they clicked on an ad or viewed a marketing article or with your product, which is more well-known, those are all events. They can all be modeled as events.

Starting point is 00:39:05 And it's super granular, super intuitive as a way to understand what people are doing. And it's really powerful because oftentimes you want to ask questions about sequences of events, right? Like which user spent this much time on my pricing page and then looked at three developer dots. That's probably a user I want to reach out to.

Starting point is 00:39:21 Or, you know, so many things can be modeled off of that. And so I think data warehouse is becoming the loading dock for all of this data, which can be very easily modeled as events. But it's not a very great analytical tool for events because SQL is optimized for rows and tables and joints and not events and sequences of events and segmentations of events. And so one of the things that we're spending a lot of time thinking about is, how do we get that really rich, trusted, comprehensive data set from the data warehouse

Starting point is 00:39:48 into a tool that's optimized from the UI down to the data model for events because that unlocks really fast, intuitive exploration of data on data set that people already have in trust. So that's, I think, one of the big trends we're excited about in what we see this future. Interesting. And you think Mix panel is in a good place to help people do that. Is that you see this? Like there's something like companies like yours will help people solve, or there's something everyone's going to have to figure out for themselves, or there's like a whole new class of startups launching to help them make the mess out of data warehouses?

Starting point is 00:40:18 No, there's always a new class of startups joining in the analytics space. It's never a dull moment. Yeah, I think this is something that we're looking to solve because, I mean, analytics is only as good as the data. And if people are already collecting great data in the warehouse, I mean, we integrated with the warehouse really well, then we get access to that good data. increasingly what we've been seeing is that companies like in the reverse ETL space like census and high touch are effectively reinventing the CDP reinventing data movement tool like segment

Starting point is 00:40:45 on top of the data warehouse. And so really our strategy there is just tightening our integration with those tools. And we've seen just huge growth and people are using their data warehouse as the source for events, not adding SDK tracking anywhere. I just saying, I already have events sitting here. I trust all of them. They're from all parts of my business. Why can't I do analytics on my support tickets and my gong calls just as easily as I can do it?

Starting point is 00:41:06 about my user behavior. And so I think that's something we're seeing and we're investing in. Awesome. Anything else you'd like to share before we get to a very exciting lightning round? Yeah, we open talking about the transition from engineering to product. And I think one of the things that's just been really fruitful in my career, both on the engineering side and products side is just adopting that product mindset and getting closer to customers, consuming the raw feed of customer context, taking every opportunity to talk to them.

Starting point is 00:41:34 And I'm really excited to see, you know, things like this podcast and your newsletter and other forums for engineers to also develop that point in my debt and get closer to customers because I think long term, that just means better products and services at lower and lower prices, which is just innovation, right? So I'm really excited to see more of that in the world. Here, here. With that, we've reached our very, very exciting lightning round. I've got six quick questions for you. I'm just going to go through them pretty fast. Whatever comes to mind, just share, and we'll see how it all goes. Sound good?

Starting point is 00:42:04 Let's do it. Okay. What are a couple books that you recommend most to other people? On the business book standpoint, there's a book called The Goal by Elihu Goldrad. And it's kind of an old book, but I like it because it's sort of written in this like fast-paced thriller model. It's like a fiction book, but it's about this idea of the theory of constraints, like finding constraints in a system and how you can remove them to improve productivity.

Starting point is 00:42:27 So I found it's just like a fun read and also really insightful. Non-technical books that I recommend to folks. particularly those that live in SF, which is a cool gray city of love by Gary Camilla, who is a long-time SF precedent, and it just goes into the history and the communities and the geography

Starting point is 00:42:45 of San Francisco, and I just discovered so many little pockets in the city from reading that book, so it's something I recommend to people who live in San Francisco. What's a favorite other podcasts that you enjoy other than this podcast? I'll do a non-tech one for this. So I'm a huge fan of the show, The West Wing,

Starting point is 00:43:01 and so there's this podcast called The West Wing, that goes to each episode of the West Wing and brings out actors from the show as well as folks from government to talk about each episode. And it's just a delight to listen to if you love the West Wing. Wait, so they go back to the old show The West Wing and talk about old episodes with politicians? Yeah. That's cool. Wow. Yeah, exactly.

Starting point is 00:43:21 So the show's over. The podcast is over. Oh, okay. You have like all seven seasons. But I think it started in 2016 or 2015 or something. Got it. So cool. Favorite recent movie or TV show that you?

Starting point is 00:43:33 you really enjoyed. Pretty mainstream TV states. So I really enjoyed a We Crushed and Severance. That's two good shows I really enjoyed last year. Awesome. Favorite interview question that you like to ask people that you're interviewing? A big side of open-ended questions. And so one of the questions I asking the behavioral interview at the start is walk me through the story of you from college to now or high school to now, if they're a more junior candidate. And a couple of interesting things here. interesting to see where people spend most of their time talking and where they don't. And also, you know, how they describe the other people on that journey. And do they use like a standard framework to describe everyone or do they, you know,

Starting point is 00:44:11 go into each person uniquely? It's just tons of follow-up questions from that. Final question is who else in the industry do you most respect as a thought leader? Got a lot of inspiration from Gibson Biddle and his product strategy, medium thing. And in particular, there's a piece of, on proxy metrics and like the shape of metrics you should use, which I found this like a really, like the way he frames it is a really elegant way

Starting point is 00:44:35 to measure a reach and impact at the same time of your metrics. And then also a big fan of Shishir from Koda, specifically as essays on on eigen questions, framing problems, and the one on marginal churn contribution. It's really interesting. Amazing. Both guests of this podcast and people I love. Great choices.

Starting point is 00:44:54 Béje, this was awesome. Thank you so much for joining me. Two final questions. Where can folks find you online if they want to reach out, learn more about what you're up to you. And then how can listeners be useful to you? I'm on Twitter and LinkedIn. I think my handles will be in the show notes.

Starting point is 00:45:09 Not super active there, but I definitely check DMs and would love to connect on either of those. And then how can listeners be useful to me? Yeah, I mean, ultimately, a Mixfinal, we're building a product for product teams. So two things, if you haven't used McFanel in the last four years, we've changed a lot as I've subscribed on the pod. So, you know, check it out and happy to take any feedback

Starting point is 00:45:29 to help us improve the products. Awesome. Bajé, thank you so much. Thank, Lenny. It's been great. Thank you so much for listening. If you found this valuable, you can subscribe to the show on Apple Podcasts, Spotify, or your favorite podcast app. Also, please consider giving us a rating or leaving a review, as that really helps other listeners find the podcast. You can find all past episodes or learn more about the show at Lenny's Podcast.com. See you in the next episode.

Lenny's Podcast: Product | Career | Growth - An inside look at Mixpanel’s product journey | Vijay Iyengar (Head of Product)

There aren't comments yet for this episode. Click on any sentence in the transcript to leave a comment.