Mixpanel Shut Down Two Products. Retention Went From 50% to 85%.

At the PDMA summit in Atlanta this October, I told someone a story about Mixpanel, the product analytics company. The way I told it, Mixpanel spent a year doing nothing but cutting features, and retention, NPS and revenue all went up. It's a great story. Going back to the sources for this post, it turns out I'd remembered it wrong.

Mixpanel didn't spend a year cutting features. It shut down two entire products, put its engineers back on the one product that mattered, and then shipped something like a hundred features in a year. The real version is better than mine, because it's more useful. It's really a story about focus, and about what focus lets a team do next.

How a successful company lost the thread

Mixpanel launched in 2009 to help product teams understand how people used their products, and it grew fast on that one idea. Around its 2014 Series B, the company decided it needed more to sell. "We sort of jumped the gun if you will, and prematurely [shifted] horizontally and started serving different markets," CEO Amir Movafaghi told TechCrunch in 2021. "We built a messaging app and we built an A/B testing experimentation product."

Each move made sense on its own. More products meant more to sell to the same customers. But Neil Rahilly, who led product and design, described the cost in the same piece: "I think we got spread out across too many different products and were trying to do too many different things for too many different customers."

By 2018, Mixpanel was losing roughly half of its customers every year. Its net promoter score, according to the company, was sitting at 15. Customers weren't leaving because they'd stopped needing product analytics. As Vijay Iyengar, Mixpanel's head of product, later explained on Lenny's Podcast, they were leaving because more focused competitors were out-building Mixpanel on its own core product.

They cut products, not features

When Movafaghi took over as CEO in 2018, the company did the hard thing. It shut down the messaging and A/B testing products entirely and refocused on product analytics, the thing it had been built to do.

Iyengar's explanation of why is the most useful part of the whole story, and it's about people, not products. "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," he said. "We moved people." And the side products weren't worth it: "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." In his telling, those adjacent products would contribute something like five to ten percent of revenue while pulling engineers away from the thing customers were actually paying for.

His rule for next time is one I'd put on a wall: "Invest profits and not people." Fund new bets with what the core earns, but don't staff them by draining the core.

Then they shipped a lot

Here's the part my conference version left out entirely. After the cut, Mixpanel didn't ship less. According to a summary of Iyengar's account, the team gathered churn reasons from customer success and sales, grouped them by problem, sorted them by how much revenue each one was costing, and made the top ten problems the roadmap. Engineers got direct access to customers with those problems. In the first year, they shipped roughly a hundred features.

That speed had a cost, and Iyengar was candid about it. Features built in a rush were siloed from each other, inconsistent, and often had to be rebuilt in different parts of the product. So Mixpanel ran a second track alongside the feature push, a design-led effort to find the fewest, most powerful building blocks the product could be organized around. At one point, leadership told the design team they'd do "zero new design projects for the next three months" so the designers could think about the system as a whole instead of getting pulled into every tactical request. Focus showed up there too, just one level down.

So the lesson I'd take from it is to stop building the wrong things, so you can build the right ones much faster. The two products they cut were taking the people they needed to fix what customers were actually leaving over.

The results showed up. By 2021, Mixpanel told TechCrunch that gross dollar retention had gone from 50% to 85%, signups were up 67% and weekly active users were up 52%. NPS climbed a long way too, although the exact figure depends on which Mixpanel executive you ask – TechCrunch reported 85, while Movafaghi later described it as rising from single digits to nearly 60. Either way, it's the same direction and a very different company than the one losing half its customers three years earlier.

The trap is people, not features

I've seen the same pattern up close. At Adaptavist, I was brought in on a product called the Library, which had cycled through three product managers and was in danger of being shut down. The team was trying to build three products at once on an architecture sized for all three combined, and every one of them was defensible on its own. Together, they needed a team five times the size. I wrote a short strategy around the one thing the Library could do that nothing else could, used it as a filter, and cut all three of the side efforts. The engineers were enormously relieved, because they finally had backing to say "not this."

Mixpanel's version is just bigger. The bigger danger for a growing company is splitting its best people across too many bets while the core starts losing to competitors who stayed focused.

I do think most products carry more features than their users can absorb. When I've looked at features against what they cost to build, how many people actually use them and how much of the team's time goes into maintaining them, a lot of them don't earn their place. But that's my own opinion from my own work, not a Mixpanel finding, and it's a smaller point than the one Mixpanel teaches.

What to look for in your own company

If you run a software company and the metric you care about won't move, it's worth asking a few questions the Mixpanel way. How many of your engineers are working on something other than the core product right now, and what share of revenue does that other thing bring in? If you sorted your churn reasons by the revenue they cost, what would the top ten be, and how many of them is anyone working on this quarter? And which of your side bets is a mild success that nobody has had the heart to end? None of those questions requires a consultant or a new framework, just a willingness to look at the answers.

Cutting a product that works is painful, and it should be. But Mixpanel's story suggests the pain of cutting it is a lot smaller than the pain of letting it quietly starve the thing your customers came for.