A Trial Frame Is Not a Prescription

An optometrist has a trick for making you doubt your own eyes. She lowers a metal frame in front of your face and drops a loose lens into the front slot. Then she asks a question with no correct answer: better, or worse? You squint at the chart. She swaps the lens for its neighbour, half a dioptre stronger. Better, or worse? You are not being tested once. You are being triangulated, one guess at a time, towards a number neither of you knew when you sat down.

That metal frame is called a trial frame. It holds nothing permanent: the lenses come out as easily as they went in, and nobody signs anything until the chart stops changing. A prescription is the opposite kind of object. It is a decision, dated and printed, that says this combination of glass is correct until further notice. Most product teams skip the trial frame entirely. They put one lens in front of the eye, call it the vision, and print it as a roadmap. Then they defend it in every meeting that follows, as though the chart had already stopped changing.

Why Locking Early Feels Like Leadership

A founder who says here is the vision gets funded. A founder who says we are still trying three lenses gets asked when the deck will be ready. Investors, engineering leads, and marketing calendars all want the same thing: a fixed target they can build a schedule against. A locked vision is cheaper to plan around than an honest one. So the room rewards conviction over correction long before anyone has tried a second lens.

That pressure has a shape in the tools themselves. A rough sketch invites debate. A polished mockup, built in the brand's real typeface with real copy in every field, reads like a decision that has already been made. Once that file lands in a board deck, reopening it costs someone's credibility, not just an afternoon. The file was built to be tested. The room treats it as signed the moment it is shared.

Prototypes Are Meant to Be Thrown Away

Jake Knapp designed the opposite discipline at Google Ventures. He called it the Design Sprint: four days building a fake good enough to put in front of five real customers on day five. The whole approach runs on one rule: a prototype is disposable, built just enough to learn and never more. A team that has not yet watched someone use the thing has no business defending it.

Here is the tension, held at full strength. The sprint's own facade can calcify the moment it survives Friday's test with a shrug instead of a verdict. Ambiguous feedback gets read as permission. The facade gets promoted to final design overnight. The team that spent four days building something disposable spends the next four months defending it, as though Friday had settled the matter for good.

What Farah Watched Get Signed Off

Farah ran her first and only round of user testing on a checkout redesign on a Thursday afternoon. Three of five users hesitated at the same step, unsure whether their new address had actually saved. She wrote it up that evening. The stakeholder deck, built from the same screens two days earlier, had already gone to the board on Tuesday. It carried a sign-off line under the mockup that nobody had thought to leave blank.

Her finding was filed as an item for version two, a version the roadmap never funded. I have done the same thing to my own testers when the room needed to move. I heard the hesitation, logged it politely, and shipped the lens I already had in the frame. Nobody lied in that meeting. Everyone had simply agreed, without saying so, that Thursday's test was theatre and Tuesday's deck was the diagnosis.

Nobody Frames the Trial Lenses

Nobody frames the trial lenses. They go back into their case within the hour: no ceremony, no signature. They are ready to be tried again on the next patient who sits in that chair. What eventually gets framed is the finished pair: engraved with a date, handed over as though the question were now closed.

Even that closed answer has an expiry. UK guidance from the College of Optometrists tells practitioners not to go longer than two years before re-examining an adult. The eye that was measured correctly in one sitting drifts, and a lens that was right then is not guaranteed to be right now. A shipped product vision rarely carries that clause. Nobody schedules the retest. The lens that was right at launch stays on the face for years after the eye it was measured for has changed.

Give the Objection Its Full Weight

Give the objection its full weight. Jeff Bezos built disagree and commit for exactly this problem. In a 2016 letter to Amazon shareholders, he described greenlighting a show he personally doubted. By that point in a genuine disagreement, he wrote, nobody can know the answer for sure. Further argument only spends time nobody gets back. A team that tries a hundred lenses and never signs a prescription has not escaped the mistake this essay describes. It has built a slower version of the same one.

Read the letter again and the difference holds. Bezos committed after a genuine disagreement had already been aired, not before anyone had looked through a second lens. Jakob Nielsen wrote the familiarity law that gets misquoted in design reviews, the one I examined in Familiarity Is a Floor, Not a Ceiling. He also found, with Thomas Landauer, that testing a design on five real users catches roughly eighty-five per cent of its usability problems. Nielsen Norman Group still publishes that finding as guidance. That is a number small enough to fit in one Thursday afternoon. Commit after that Thursday and you are disagreeing and committing. Commit before it and you are writing a prescription off the first lens you happened to try.

The Lens Nobody Swapped

A trial frame lets you find out which lens is correct before anyone signs anything. A prescription is what you write after that finding, not a substitute for having looked. Farah's three hesitant users are still squinting at the first lens nobody let her swap.