A user opens a new tool, pokes around for seven minutes, and closes the tab. No unsubscribe, no complaint, no support ticket. Product teams log it as a churn event. The inference that follows is that the user tried the product and decided it wasn't worth continuing. That reading is almost certainly wrong.

The question during those seven minutes wasn't whether the product was useful; they couldn't evaluate that yet. It was whether the early experience was worth continuing. The churn event was a verdict on the first session, not the product. That distinction has a different explanation and a different solution, and confusing them is expensive.

Beyond utility

The conventional explanations for this kind of exit are well developed. Maybe the product isn't right for the user's job. Maybe onboarding failed to deliver a moment of value before the user ran out of patience. These diagnoses are reasonable, and each has generated substantial investment: activation metrics, product tours, value demonstration flows. They address a real problem, and that problem is utility: making the product's value visible early enough for the user to recognize it.

Utility provides instrumental motivation: the user can see the value ahead and is willing to invest in reaching it. But instrumental motivation requires outcomes the user has not yet experienced, and in early use, there are none. The unmeasured variable is the intrinsic motivational cost of learning. This is distinct from ease of use, which is a property of the path from input to output, and from computer anxiety, a stable individual trait. Motivational cost is a property of how the path feels to travel. The cost builds session by session, accumulating from whether errors seem recoverable, whether progress feels legible without being labeled, and whether exploration seems worth the effort. Poor early sessions compound.

Apple's Lisa team recognized this in 1979. The formal product specification listed fun to use as a design goal alongside ease of use and minimal training. The rationale was that Lisa would not survive as a system people used because the boss told them to; without a company mandate, the experience itself had to give people reason to keep coming back. The Lisa shipped in 1983 and its interface became the direct ancestor of the Macintosh. The requirement written into that spec is the missing context for a problem product teams often misread.

Two adoption contexts

How much the motivational cost matters is determined by two structural properties of the adoption context. The first is institutional embeddedness: how much organizational forces insulate a product from user disengagement. Mandate, switching costs, network effects, and organizational inertia all contribute. An enterprise resource planning system doesn't need users to enjoy learning it. Organizations pay the motivational cost through training budgets and implementation consulting, not through the user's own willingness to continue.

The second is discovery path length: how long from first contact to a genuine experience of the product's value. For some products this path is short. A calendar app demonstrates its value in minutes. For others, value only appears after sustained investment, and the user decides repeatedly to continue without evidence that continuing is worthwhile. Feature completeness matters less in these contexts than it appears to, because reaching the features requires surviving the learning phase.

These two variables interact. A mandated enterprise tool with a long discovery path survives poor learning-phase design because the institutional context absorbs the motivational cost. In low-embeddedness contexts with long discovery paths, a rewarding early experience matters more to adoption than a complete feature set. Product teams frequently prioritize the latter when they should be attending to the former. AI-native work tools are the sharpest instance. Their value is not a fixed feature set a product tour can demonstrate; it emerges from how users learn to engage with a generative system whose capabilities only reveal themselves through use. The discovery path is open-ended, and value is invisible in a feature list and undemonstrable in any demo.

Three conditions

Fun in early use is the feeling that a product is starting to make sense. Researchers studying intrinsic motivation and skill acquisition have identified three structural conditions that determine whether that feeling takes hold. The product must have depth worth modeling; sessions must yield signal that the model is improving; and the system's rules must be stable enough to generalize.

The first is discoverable depth. A product has it when exploration has positive expected value, when poking around reveals something the user did not know was there. The motivation to explore arises from a perceived gap between what is known and what could be known; that gap must be sensed before it can drive behavior (Loewenstein, 1994). A product creates that gap by offering the sense that exploration will pay off and that there are skills worth developing, doing the motivational work before users have real outcomes to evaluate (Hassenzahl, 2004). Products with discoverable depth leave users with the sense that more remains to find. A product without it closes the information gap within a few sessions, and exploration stops.

The second is prediction improvement in the sense that the user feels they can better anticipate how the product will behave. A session produces prediction improvement when the user finishes with a clearer mental model of the product's logic. They can more reliably slot it into their existing understanding of their work, and fewer interactions surprise them. This is measured not by features completed but by whether the user's expectations are more often confirmed than they were before. Learning progress, the rate at which a mental model's accuracy is improving, is the primary driver of intrinsic motivation in early skill acquisition (Oudeyer et al., 2016). This reward does not require external confirmation; the user feels it in the growing accuracy of their expectations. A session that leaves the user surprised by the same things as the previous session has failed this test. Each session either improves the model or compounds the cost of continuing.

The third is coherent logic. A coherent model is only possible if the system's rules are stable. A product has coherent logic when understanding one part predicts the behavior of others, when the rules the user infers from early interactions continue to hold as they go deeper. A mental model becomes useful only when a user's representation of a system shifts from a collection of observed behaviors to a predictive theory (Norman, 2004); coherent logic is the structural condition that makes this shift possible. Products without coherent logic require users to rebuild their model with each new feature, and each rebuild carries the full motivational cost of the learning phase. Coherent logic is the structural property that makes accumulated understanding possible.

The design problem

The 7-minute user who closed the tab didn't decide the product was limited. The product's value was latent, the discovery path was long, and nothing in the early experience gave them reason to continue. Product teams had no way to distinguish that from a user who had tried and rejected the product. The exit registered as churn. The real event was a quiet failure of learning-phase design.

Each condition points to a different design problem. Discoverable depth is a structural property of the product's information architecture; it requires that exploration rewards curiosity rather than exhausting it. Prediction improvement is produced by sessions that consistently close the gap between what the user expects and what the product does, giving the user evidence that their model is improving. Coherent logic requires that the rules the user infers early on continue to hold as they go deeper, rather than treating every new feature as a fresh design problem. Each property requires design effort directed at the felt quality of early interactions, not at features and not at onboarding flows. These conditions operate below the reflective level; they are felt before they are interpreted, and they do not habituate.

Product teams routinely ask whether their product works. For products where value is latent and the discovery path is long, the prior question is whether getting good at it was designed with the same care as the features themselves. The voluntary market pays the motivational cost user by user, session by session, in quiet exits that generate no tickets and no complaints. Organizations absorb that cost through training budgets and mandatory adoption. Individual users, given no such buffer, make a different calculation.

References

Hassenzahl, M. (2004). The interplay of beauty, goodness, and usability in interactive products. Human-Computer Interaction, 19(4), 319–349.

Loewenstein, G. (1994). The psychology of curiosity: A review and reinterpretation. Psychological Bulletin, 116(1), 75–100.

Norman, D. A. (2004). Emotional design: Why we love (or hate) everyday things. Basic Books.

Oudeyer, P.-Y., Gottlieb, J., & Lopes, M. (2016). Intrinsic motivation, curiosity, and learning: Theory and applications in educational technologies. Progress in Brain Research, 229, 257–284.