Tracking links that survive a rotation

The problem with rounding

Ask ten people whether they’d tap an icon and you’ll get ten different kinds of yes. Some mean it. Some are being polite. Some are thinking about it. A scale that only lets them pick a 4 or a 5 treats all of those the same.

That matters most when you’re choosing between options that are close. If two icons both get a lot of 4s, a whole-number scale calls it a tie, even when one of them got the more convinced 4s.

What we keep instead

Personas answer in their own words, so the hesitation is right there in what they say. We keep it, and the score lands between the points of your scale rather than being rounded onto one of them.

“I’d probably tap it” and “I’d tap it straight away” are not the same answer.

Where it changes the result

The difference shows up when you rank options. Two options that look tied on a whole-number scale often separate once the strength of each answer is kept.

What to do about it

Write your five statements so each one sounds different from its neighbours, and keep the wording of the question the same across every option you test. Then let the panel tell you how sure they are.