A prototype is not a miniature product. It is a decision tool — something you build specifically so an assumption you cannot see stops being invisible. The moment a user fumbles with a control, hesitates before confirming an action, or simply cannot find the thing you assumed was obvious, the assumption becomes a fact you can act on. The hard part isn’t building that moment. It’s knowing what to do the morning after.
A test just wrapped. The numbers are in front of the team, someone opens with “well, this basically confirms what we thought,” and within ten minutes the meeting has quietly turned into a defense of the idea everyone liked before the test even started. Nobody voted for that outcome. It just happened — because the human brain is built to skip the hard part of deciding and go straight to the comfortable part of agreeing.
You’ve probably run the numbers already. A mid-level IT hire costs somewhere in the £35,000–£45,000 range annually, and the managed service provider quote sitting in your inbox looks steeper on a monthly basis. Case closed, right? Except that comparison is a bit like comparing the price of a car to the price of a taxi ride and concluding the car is always cheaper — you’ve only counted the fuel, not the insurance, the repairs, the parking, or the years it sits idle in the driveway. When founders and small teams make IT staffing decisions this way, they’re not comparing costs. They’re comparing two different categories of financial and operational risk and pretending they’re the same number.
A founder runs a two-week test on a new onboarding flow. The numbers come back mixed — engagement up slightly, but a handful of users seem confused. In the debrief, one product manager wants to ship it. Another has doubts but says nothing, because the last time she pushed back, the conversation got tense and nothing changed anyway. The team ships. Three months later, they’re trying to figure out why adoption stalled — and nobody remembers that someone already saw it coming.
A founder message a colleague forwarded me recently read something like: “Everyone on the leadership team wants an answer by Friday — do we buy the AI thing or not?” The question sounds simple. It isn’t. Buried inside “should we buy AI” are at least three separate decisions: how much money to risk, how fast to move, and how prepared your organization actually is to use what it buys. Treating those as one binary choice — all in or nothing — is how good teams end up either overpaying for tools they outgrow in months or watching competitors pull ahead while they wait for a “clear winner” that may never arrive.
When an early product test comes back with a promising number — a waitlist that fills up, a pilot customer who says yes, a demo that gets forwarded around — most teams ask one question: is this signal real? Almost nobody asks the second, equally important question: are we the kind of team that should trust this signal enough to act on it?
