Phil Libin came through Draper last month. Evernote CEO for years, now doing other things. The founders in the room knew Evernote. Most of them use it or used it. What they didn't know was the thinking behind it.
The talk wasn't what I expected. He didn't do the standard origin story. He spent most of the time on one question: what makes people come back to something ten years later.
The retention problem nobody solves
Growth is relatively well understood. You can optimize acquisition, you can A/B test onboarding, you can run referral programs. These are engineering problems with known approaches. Long-term retention is a different category.
He said that most products built around daily active users are optimizing for a metric that has nothing to do with whether the product is actually good. A product people open every day out of habit is not the same as a product people value. You can build the former without building the latter. The former is easier and the metrics look better in the short run.
Evernote's retention thesis was different: people don't open Evernote every day, but they never leave, because over time it becomes the place where their thinking lives. The value compound over years. A note from 2012 is still there in 2019. The longer you use it, the harder it is to replace.
He called this asymmetric retention. The product gets stickier the longer you use it, not because of switching costs in the hostile sense, but because the value is genuinely accumulating. Each year of use makes the product more valuable to that specific user.
The thing about taste
He was direct about something I've heard from other founders but rarely stated this plainly: most feature decisions at Evernote were made on taste. Not on data. Data would tell you whether a feature was being used. Taste was the question of whether it should exist.
The tension between taste and data is real. Data tells you what users do. It doesn't tell you what they would do if you'd built something better. Products that are only data-driven regress to the median of existing behavior. Products that ignore data ship things nobody wants. The difficult skill is using data to validate taste, not to replace it.
He said they killed features regularly that had usage data justifying their existence, because the features made the product feel cluttered in a way the usage numbers couldn't see. This is a hard call to make. Usage exists. Removing something with usage makes users unhappy. The metric-optimizing argument for keeping it is always available.
What the founders in the room did with this
Most of them were building for growth. Their investors want growth. Their metrics are growth metrics. Thinking about whether someone will still use their product in ten years is not the immediate pressure.
But the question stuck with several of them in ways I noticed in later sessions. One founder building a productivity tool for remote teams started asking whether his features were building long-term value or just adding engagement. That's a meaningful shift in how you evaluate a roadmap.
What I took from it
Asymmetric retention is a useful frame for evaluating product decisions even at early stage. Does this feature make the product more valuable the longer someone uses it, or does it just make it more engaging today. The second is easier to build and harder to sustain.
The products I've respected most over a long period of time are ones where I discovered new value after years of use. That's rare. It's also the thing worth trying to build.
With gusto, Fatih.