What people say, what people do

Users are the experts of their own experience. Most of that expertise sits below the surface.

A room agrees, then walks out

Predict before you look

The story goes that Sony’s researchers gathered a focus group to settle a colour for a new boom box: yellow or black. The room agreed on yellow. Fresher, more modern. On the way out, the facilitator pointed to two stacks of free boom boxes by the door, one yellow and one black. Theirs to take home.

Which did they take?

Sanders and Stappers call this the say/do gap. What people say differs from what they do. Sometimes it is to please the interviewer. Sometimes it is to appear competent. Often they have simply forgotten the details that matter most.

This is what is behind the line “users don’t know what they want”. It is half right. They know a great deal. The job is bringing the right tools to reach it.

Four levels of knowledge, each harder to reach

The surface

Explicit what people say

Knowledge that can be put into words. Goals, frustrations, opinions, preferences.

Reached byInterviews and surveys

Observable what people do

Behaviour others can see, which the person often doesn’t notice in themselves.

Reached byWatching, in context

Tacit what people know but can’t say

You can open a website on your phone with your hands full. Explaining how would be hard. Tacit knowledge shows up in the things people make to get by.

Reached byArtefacts and workarounds

Latent what people haven’t met yet

Needs not yet experienced, but felt through past ones. Before ride-hailing apps, nobody asked for live tracking. They asked for certainty about waiting, paying and pickup.

Reached byMaking things together

Reaching each level takes a different move

Start with how you ask, not only what you ask

A student says the quiz was confusing. The instinct is to ask why. “Why” invites a tidy explanation built after the fact. Anchor in a specific moment instead.

× Invites a story

“Why was it confusing?”

✓ Anchors a moment

When did you last try it? Where were you?
What else were you doing at the time?
What happened right before you stopped?

When you watch, write down what happened, not what it means

Records survive disagreement. Conclusions start it. Sort these field notes.

Record, or conclusion?

Student clicked Submit twice in 3 seconds.
Student was confused by the submit flow.
Teacher opened the class list in a second tab before marking.
Teacher didn’t trust the auto-save.
“I always print this page first,” she said, holding a printout.

Look for what people have built around the system

People rarely register their own artefacts as workarounds. The laminated note is just how you do it.
Direct questions rarely surface them. Anchoring in a moment, “walk me through what you did yesterday”, can get close.
The most reliable way in is being in the room where the work happens, and looking for what has been taped up, printed out or kept open alongside the system.

The spreadsheet beside the platform is the finding.

Give people something to react to, or to build

Not “what do you need from assignment tracking?” but “here’s a sketch of how this could work: what would it change for you?” Better still, give people materials to construct a flow rather than describe one: paper cut-outs, screen sketches, building blocks.

Asked to arrange how she would like to check her homework, one student set it up so her mother saw the results first. That is the insight that counts. At home, homework review is something they do together. No interview question would have found it.

Take this with you

What people say is the top layer. Below it sit what they do, what they know but can’t say, and what they haven’t met yet.
Ask about a specific moment, not “why”. Write records, not conclusions.
Look for the workarounds people have built. Then give them something to react to or build.

If your research only asks, you are designing for the top layer.

Try this week

Go one layer deeper than last time.

Take your next interview guide. Rewrite every “why” as a question about a specific moment.
Book thirty minutes to watch the work where it happens. Photograph anything taped, printed or kept open beside your product (with permission).
In your notes, underline every word about someone’s inner state. Replace it with what you saw.

To mull over

What workaround do you use every day without thinking of it as one?
Which of your product’s features was built from what people said, and how would it look if it had been built from what they did?
Whose experience of your product happens with someone else in the room?

What people say, what people do.