Useful Friction

The default religion of startups is removing friction. Every pitch deck, every product review, every design critique eventually arrives at the same commandment: make it easier. Reduce steps. Eliminate barriers. Smooth the path.

And most of the time this is right. Bad friction is real. Confusing checkout flows kill conversions. Slow load times kill retention. Nobody ever won a market by making their product harder to use for no reason.

But there's a second kind of friction that gets lumped in with the first, and I think that's a mistake. Because some of the most defensible companies in the world got that way not by removing friction, but by adding it. On purpose. And the friction is the thing that makes them impossible to leave.

Duolingo is the clearest example I know of.

Language learning apps have some of the worst retention numbers in software. The average learning app retains about 1.76% of users. That's almost nobody. People download the app, try a lesson, and disappear. In 2012, Duolingo's next-day retention was 12%. Better than average but still terrible for a product that only works if you use it every day for months.

Their solution wasn't to make the product easier. It was to make it stickier. They built the streak.

The streak is a simple counter. It tracks how many consecutive days you've completed a lesson. That's it. There's no technical innovation here. It's just a number that goes up if you show up and resets to zero if you don't.

And it changed everything. Users who hit a 7-day streak became 2.4 times more likely to come back the next day. When Duolingo added a streak widget to the iPhone home screen, daily commitment jumped 60%. They kept tweaking the feature, adding streak freezes and wagers and milestone animations, and by 2023 they hit 17 million daily active users. In an app category where almost nobody comes back.

The interesting thing about the streak is that, strictly speaking, it makes the product worse. It adds pressure. It creates anxiety. People report opening Duolingo at 11:59pm in a panic, not because they want to learn Spanish but because they can't bear to lose a 200-day streak. There are people who maintained their streaks on their wedding day. During childbirth. This is not the behavior of someone enjoying a frictionless experience.

But here's what I think most people miss. The streak isn't a feature. It's a switching cost. Every day you maintain your streak, you're building something you can't take with you. If you switch to Babbel or Rosetta Stone or any competitor, your streak goes to zero. Three hundred days of consistency, gone. The counter doesn't transfer. And because losing a streak feels like losing something real (even though it's just a number on a screen), people stay. Not because Duolingo is the best language app. Because their streak is inside Duolingo.

This is the pattern I want to name. Useful friction. It's friction that doesn't make the product harder to use. It makes the product harder to leave.

There's a version of this in enterprise software that's even more extreme, and I think it explains something important about why Salesforce is still dominant despite everyone complaining about it.

Implementing Salesforce costs somewhere between $25,000 and $500,000, depending on the size of the company. That's before you pay for the software itself. The implementation involves consultants, data migration, custom workflows, integrations with other systems, and months of training. A mid-market company might spend $75,000 to $150,000 just getting the thing set up.

This is, by any normal definition, terrible friction. It takes months. It's expensive. It's painful. And Salesforce knows this. They could make implementation easier. They've had decades to simplify the onboarding process.

But think about what happens after a company finishes a six-month, $100,000 Salesforce implementation. They now have every customer relationship, every deal, every pipeline, every workflow encoded in this system. Their sales team has been trained on it. Their reports depend on it. Their integrations pipe data into it from a dozen other tools.

Now a competitor comes along and says, "We're 30% cheaper and our UI is cleaner." The CTO looks at the competitor, looks at the six months and $100,000 already spent on Salesforce, and says, "We're not switching." Not because Salesforce is better. Because the cost of switching is higher than the cost of staying.

The implementation isn't a bug. It's the moat. Every dollar and every hour a customer spends setting up Salesforce makes it harder for them to leave. Salesforce's entire business model is built on the idea that the most expensive onboarding in enterprise software is also the stickiest. And the numbers bear this out. They've been the dominant CRM for over two decades despite having a product that almost everyone admits is clunky and overpriced. The friction is the strategy.

Slack does something similar but more subtly. When a company adopts Slack, they set up channels, establish naming conventions, build integrations, pin important messages, create workflows. Over months and years, the workspace becomes the institutional memory of the company. Conversations, decisions, links, files, context. All of it lives in Slack.

Then Microsoft Teams shows up and says, "We're free with your Office 365 subscription." And the company considers switching, but then someone asks: what happens to two years of searchable conversations? The tribal knowledge embedded in thousands of channels? The integrations we built? And the answer is: it doesn't transfer cleanly. The switching cost isn't the price of the software. It's the accumulated investment in the workspace itself.

Every message sent in Slack makes Slack harder to leave. That's useful friction.

I think what's interesting about all of these examples is that the friction doesn't look like friction from the inside. When you're maintaining a Duolingo streak, it feels like personal discipline. When you're customizing your Salesforce instance, it feels like doing your job. When you're building Slack channels and workflows, it feels like organizing your team.

None of it feels like you're being locked in. But you are.

This is the difference between hostile friction and useful friction, and I think most founders get it wrong because they only think about the first kind. Hostile friction is friction that serves nobody. A confusing interface. A slow loading screen. A signup flow that asks for your phone number before showing you the product. This kind of friction is always bad. Remove it.

Useful friction is friction that serves the user in the moment but serves the company in the long run. The streak helps you build a habit. It also makes you a captive customer. The Salesforce implementation helps you run your business. It also makes you unable to leave. The Slack workspace helps you communicate. It also makes switching catastrophically expensive.

The user gets value from the friction. That's what makes it useful. But the company gets something even more valuable: a moat that the customer builds for themselves.

I think this also explains something about why certain products feel "right" even though they're objectively harder to use. People love Notion, but Notion is much harder to set up than Google Docs. People love Figma, but Figma requires more upfront investment than Canva. People love Linux, but Linux requires orders of magnitude more effort than macOS.

In each case, the effort creates ownership. And ownership creates loyalty. And loyalty creates a moat. The causal chain goes: friction to investment, investment to attachment, attachment to retention.

If you told a startup founder that their best strategy for retention was to make their product harder to set up, they'd think you were crazy. But that's essentially what the most defensible companies in the world have done. Not harder to use. Harder to have used. There's a difference. The friction comes at the beginning, or it accumulates gradually. But by the time you notice it, you've already built something inside the product that you don't want to abandon.

The conventional wisdom is that the best products are the ones with the least friction. I think the more accurate version is that the best products are the ones that move friction from the wrong places to the right places. They make it easy to start doing something and hard to stop.

And the really clever ones make the friction feel like a feature. So the customer never realizes the walls they're building around themselves are the walls the company wanted all along.