Building a Product People Come Back To

A retention curve that flattens is rarely luck. Underneath it is almost always a habit, deliberately engineered into the product so that coming back stops being a decision and becomes a reflex. The return visit is the whole game in consumer, and habits, it turns out, have a structure you can design for. The same structure also marks a line every founder eventually has to stand on: the difference between forming a habit that serves the user and exploiting one that does not.

Retention is a habit, and habits have a structure

In The Power of Habit, Charles Duhigg described habits as a loop of cue, routine and reward, repeated until the cue alone triggers the routine. Nir Eyal adapted this for products in the Hook model, a four-step cycle that, run enough times, makes a product the default response to an internal itch. The point is not novelty but frequency: a product becomes a habit when the loop closes often enough that the user stops deciding to use it.

Variable reward is the engine

The step that does the heavy lifting is the variable reward. A reward that is perfectly predictable stops being interesting once the novelty fades; a reward whose timing or content varies keeps attention engaged, the same mechanism behind why a feed you scroll or a message that might be waiting pulls so hard. This is powerful and double-edged. Used to deliver genuine value the user came for, variability deepens a good habit; used to dispense empty novelty, it manufactures compulsion with no payoff, and that gap is exactly where the ethics live.

Investment is what makes it stick

The fourth step is the one founders most often skip, and it is what separates a fad from a habit. When a user puts something into a product, a playlist, a profile, their history, a network of friends, two things happen: the product gets better for them specifically, and leaving now means abandoning what they built. That stored value is also the honest version of a switching cost, the kind earned by accumulated benefit rather than by locking the exit. Every lap of the loop should leave the user a little more invested, so the product they return to next week is better than the one they used this week.

The ethical line, which is also the durable one

Engineering habits invites an obvious worry, and it should. Eyal's own answer is a simple test he calls the manipulation matrix: build it only if it materially improves the user's life and you would happily use it yourself. A habit that genuinely serves the user compounds into loyalty; one that extracts attention without giving value back is a dark pattern, and dark patterns are a short-term tactic with a long-term cost. Compulsion without value shows up later as the very churn the loop was meant to prevent, because resentment is not a foundation for a flat retention curve. The engineerable habit and the ethical habit turn out to be the same habit.

How we read it

Looking at a consumer product, we try to name the loop out loud: what is the internal trigger, how small is the action, where is the variable reward, and what does the user invest that improves the next visit. A product with a clear, value-giving loop tends to show it in the cohort curve we care about in reading the cohort curve. We are also wary of engagement that looks extractive, because the metrics can glow for a while before the trust runs out. The habits that hold are the ones a user would defend if you tried to take the product away, and that is lever three, retention, doing its real work in the four levers.

The takeaway

People come back to products that have become a habit, and habits are built, not hoped for: a trigger that fires on a real feeling, an action made effortless, a reward worth returning for, and an investment that makes tomorrow's visit better than today's. Build that loop around genuine value and it compounds into the flat retention curve every other lever depends on. Build it around empty compulsion and it borrows growth from a future that eventually sends the bill.