Focus group test, watched.
Also called: Group product test · Reaction session · Small-panel review · User group session
Putting the real, working prototype in front of a small group and reading what their hands do, not just what their mouths say.
A real object lies less than a concept, but a group still flatters the maker. Design the session against politeness: watch behaviour, not just words, run small panels, and trust the hands over the nodding. There’s a worked test below, and a checklist.
What a focus group test is
A focus group test puts the actual prototype, the working box or the looks-like-finished shell, into the hands of a small group of target users and watches what happens. Not a slide of the idea. Not a description read aloud. The real object, on the table, switched on, with a price tag attached. A concept invites people to imagine; an object forces them to react.
That reality is the whole value, and it’s also where the trap hides. An object lies less than a concept, true. But a group of people sitting in a room still wants to be kind to the person who built the thing, still defers to whoever spoke first, and still says the encouraging thing because disagreeing out loud feels rude. Politeness is the default setting of a group, and politeness is what kills a product quietly six months later when nobody actually buys it.
So the session has to be engineered against its own social grain. The job is not to collect compliments. The job is to read the gap between what people say and what their hands do, because the hands are honest when the mouth is being polite.
Where groups flatter you
- The first speaker sets the room. One confident “I love it” and the rest of the group calibrates to agreement. Run the reactions privately before anyone speaks aloud, or the first voice becomes the only voice.
- The maker in the room is a thumb on the scale. People soften everything when the person who built it is watching them hopefully. I learned to shut up, sit back, and let a neutral facilitator run it.
- “Would you buy it?” is the worst question you can ask. Everyone says yes to a hypothetical. Watch instead for whether they actually reach for their wallet, or flinch at the price, or put the object down and don’t pick it up again.
A worked focus test
We pushed the proofing box across a table at three small panels of Sourdough School bakers and mostly kept our mouths shut. Here is what that session looked like, so you can see the shape of a real test rather than a generic template.
Notice the gap. Spoken, the price was “a lot for a box”. Behaviourally, not one baker stopped handling it, and two asked when they could buy one. We weighted the hands over the words, and the £149 band held.
- Ask the group “do you like it?” and write down the yeses.
- Let the loudest, most confident voice speak first.
- Sit in the room beaming while they hold your prototype.
- Record the compliments and call it validation.
- Capture each reaction privately before anyone talks.
- Read the hands, the face at the price, the unprompted reach.
- Get the maker out; let a neutral facilitator run it.
- Weight what they do above what they say.
How it fits the bigger picture
Focus group test is activity 08.20.01 in the framework, an early Develop-stage test. It feeds straight into the A/B test (08.20.02), which takes the questions a group can only hint at and answers them cleanly by pitting two versions against each other on real choices.
What it can do
It surfaces the gap between what people say and what they do, while the product is still cheap to change. A real object in real hands tells you in an afternoon what a survey never will: whether the price stings, whether the no-app stance lands, whether anyone reaches for it without being asked.
What it can’t do
It can’t give you a clean number. A group is small, social, and swayed by whoever spoke first, so its output is a strong signal and a set of sharper questions, not a statistic. For that you run the A/B test next, which isolates one variable at a time.
See the full 10-stage process →
Try it yourself
Take your most finished prototype and find four or five real target users, not friends. Put the object on the table and stop talking. Have them write a private first reaction before anyone speaks. Watch their hands. Ask what they’d expect to pay before you reveal the price, and read the face when you do. Never ask “would you buy it?”
Want help framing the test before you run it? Start the Free Sprint → and the GPT will help you sharpen what you’re actually trying to learn.
Your focus-test checklist
Project notes: the hands disagreed with the room
▸ From the notebook · optional reading
Three small panels of Sourdough School bakers, a £149 box on the table, and the day we stopped trusting what the room said out loud.
3 min read · click to open
Dan and Anna Hartley wanted to run one big group of a dozen bakers in Stockport. We pushed back: three small panels of four or five instead, run separately, because a dozen people in a room produces a consensus, not data. They agreed, reluctantly.
The first session went wrong, usefully
Anna sat in on the first panel, beaming, holding the box up like a proud parent. Every baker was lovely about it. “Beautiful.” “I’d definitely have one.” We had a page of compliments and almost nothing real. So for panels two and three I asked Anna and Dan to leave, brought in a neutral facilitator, and changed the running order.
Each baker wrote a private first reaction on a card before anyone spoke. Then they handled the box with no prompt while the facilitator just watched. Only then did we talk, and we never once asked “would you buy it?”
What the hands said
Out loud, the verdict was that £149 was “a lot for a box you can’t even control from your phone”. On the cards and in the handling, a different story. Nobody put it down. Two bakers asked, unprompted, when it would be on sale. When the facilitator mentioned there was no app, the room visibly relaxed, one baker said “good, I’ve got enough things nagging me from my phone”.
The spoken price objection was real but soft, the kind of thing people say about anything over a hundred pounds. The behaviour, the stroking of the wood, the unprompted reach, the relief at no app, told us the £149 no-app box had a genuine audience. We carried both findings forward: hold the price, hold the no-app stance, and sharpen the value story so the spoken objection had a clean answer.
— Develop stage, project notes, 2026
— Next in Develop → A/B test