Picture a founder with a promising app and a choice to make. She could quietly test the product with a handful of customers, learning whether people actually want it while keeping the whole effort under wraps. Or she could stage a public launch, complete with press coverage and a visible waitlist, generating sharper feedback and maybe attracting investors. The second option teaches her more, faster. It also lets potential rivals watch closely, and what they see is not just how customers respond. They also see that she was confident enough to go public in the first place.
That second signal, the one buried in the choice itself, sits at the center of a working paper from the National Bureau of Economic Research. The author argues that entrepreneurial experiments are not only private learning exercises. They are public acts that competitors interpret, and that interpretation carries a real strategic price.
A question about visible experiments
The study comes from Joshua S. Gans of the University of Toronto’s Rotman School of Management. His starting question is straightforward: how should an entrepreneur experiment when the experiments themselves are visible? Most research on startup experimentation treats it as a matter of learning under uncertainty. Run a test, gather a signal, update your beliefs, decide what to do next. Gans adds a wrinkle that he says changes the calculation. In many real settings, the same action that produces information also broadcasts information to outsiders.
A closed beta, a conference demo, an open waitlist, a public usage milestone: each one generates evidence for the founder and is itself read as evidence by everyone watching. The decision to launch loudly can reveal that the founder is optimistic, even before any customer data arrives. Gans’s aim is to build a framework that separates these two kinds of leakage and traces how they shape what founders do.
Building the model
Because this is a theoretical paper, the method is a mathematical model rather than a survey or experiment. Gans constructs a setup with a few moving parts. A venture is either viable or not. The founder privately holds evidence about which is true, and in each period she picks one of four moves: abandon the project, run a low-publicity test (think a quiet pilot), run a high-publicity test (think a public launch), or scale up the business.
Each experiment produces two signals. One is private, observed only by the founder, and updates her own confidence. The other is public, a noisier version of the same evidence that outsiders can see. The model treats the public signal as a blurred copy of the private one, so the founder never learns less than the outside world; she simply learns it more clearly. Meanwhile, a potential rival watches the founder’s moves and the public results, and decides whether to jump in and copy the idea. An investor sits in the background too, willing to fund a scale-up only once the founder’s evidence is strong enough.
The organizing idea is what Gans calls competitive exposure. He defines it as the gap between how much the rival currently believes in the venture and the level of belief at which entering the market becomes attractive. He describes exposure as a budget: the public runway a founder has before a competitor decides it is worth pouncing. A founder with lots of exposure can afford more visible activity. A founder with little exposure risks triggering imitation with even a small public move.
Two ways the budget gets spent
The heart of the paper is a kind of accounting system for how that budget drains away. Gans identifies two distinct forces.
The first he calls leakage burn. This is the familiar one. When a public test produces favorable results, outsiders see those results and grow more confident the venture will work, which uses up exposure. The second force is the paper’s main contribution, and Gans calls it action burn. This is exposure consumed by the choice of experiment itself, before any outcome appears. If only confident founders tend to launch publicly, then the mere act of launching publicly tells rivals something. The move leaks confidence regardless of how it turns out.
Gans sums this up in a simple rule of motion: a venture’s next level of exposure equals its current level, minus action burn, minus leakage burn. He describes this equation as “the spine of the paper.”
One consequence is that a founder’s current public reputation, captured as a single number, is not enough to predict what happens next. Gans shows that two ventures can leave rivals with identical beliefs today, yet a future move can mean very different things in each case, depending on which kinds of founders were behind each history. The composition of who might be in the running matters, not just the average level of optimism.
What the analysis suggests
Working through the model, Gans draws out several patterns. The first concerns how founders sort themselves across the two types of experiments. He finds that founders with weaker private evidence tend to prefer the quiet test, because their main reason to keep experimenting is to preserve the option to learn more later, and a stealthy move conserves their exposure budget. Founders with stronger private evidence are more willing to accept a public test, because faster learning is worth more to them and they can better absorb the exposure cost of being seen.
Gans argues this produces a stealth-to-traction pattern that is not imposed by assumption but emerges naturally as different types of founders make different choices. More public runway, he finds, tends to push founders toward visible experiments earlier, while founders close to the danger zone lean on quiet ones.
A second pattern concerns scaling up. One might expect competition to push founders to commercialize sooner. Gans argues this happens only on a specific margin: when competition erodes the value of waiting more than it erodes the value of scaling. Otherwise, competition may change how a founder experiments, nudging her toward quieter pilots and smaller test groups, without making her scale any earlier. He frames this as a reason that looking only at launch dates or funding dates could miss the real effect of competition, which may show up first in the quiet design of experiments.
A third pattern involves financing. In the model, an investor’s funding standard acts as a gate that scaling must pass. When that standard gets stricter, fewer ventures can scale, so the act of scaling becomes a more selective and more informative signal to rivals. Gans interprets this as financing affecting competition not through the cash itself, but through what a funding-gated scale-up tells competitors about the venture’s private evidence.
What founders might take from it
The practical thread Gans pulls is that founders are effectively budgeting exposure whenever they experiment, publicize, raise money, or scale. The relevant question, in his framing, is not whether visibility is good or bad in general, but whether a given public move buys enough private learning or strategic advantage to justify the runway it consumes.
Some caveats are worth keeping in view. This is a theoretical model, not an empirical study, and it has not been peer reviewed. Its conclusions follow from its assumptions: a binary viable-or-not state, a short-lived rival in the baseline case, and a stripped-down treatment of financing. Gans is explicit that these simplifications are meant to keep the central mechanism transparent rather than to capture every real-world detail. He devotes a section to how researchers might eventually test the ideas, noting the difficulty of separating action burn from leakage burn in actual data, since a public launch with a good outcome bundles both effects together. For now, the contribution is a way of thinking about a familiar founder’s dilemma, reframed as a question of public action under a watchful eye.
