How it works
From question to proof, in one place.
Install once, ask a question, see what people saw and missed, ask them why, ship the fix, and know whether it worked.
Sign-up opens shortly.
Six steps
What happens, in order.
- 01
Install
One line on your site, or the Shopify app. Keep Clarity if you like.
- One line, or the Shopify app Building
- Runs beside Clarity Building
- 02
Ask
Pick a question or let us suggest one: "Why do mobile visitors leave the product page?"
- Ask a question, or let us suggest one Building
- 03
See
Per element: seen, ignored, missed and hovered. Attention where they really looked. The 3–5 moments worth watching.
- For every element on every page: seen, ignored, hovered, missed, by device Building
- Attention counted only where people really looked Building
- The 3–5 moments worth watching, at the exact second Building
- 04
Ask them
Run a study with your own customers. The interviewer knows what each one did and missed.
- Studies with your own customers, through your own link Live
- An interviewer briefed on what each person did and missed Building
- 05
Fix
The change goes to your coding agent, developer or agency.
- Your coding agent gets the fixes Building
- 06
Know
Checked against your other pages: worked, didn't, or can't tell yet.
- Every change checked against your other pages: worked, didn't, or can't tell yet Building
See what they saw
Every element, on every page, without mapping.
RevealWhy is being built to count an element as seen when at least half of it was on screen for at least half a second, and from that to report what was seen, ignored, hovered and missed, by device, for every element, with no zones to draw and no mappings to keep up. What's live today is tagged on the right.
- For every element on every page: seen, ignored, hovered, missed, by device Building
- No zones or mappings to build Building
- What visitors never saw, with where to move it Building
- Useful, distracting or missed, per content element Building
- Content over time and by segment, with intervals Building
- Recordings and heatmaps included Live
How we measure attention
Nothing gets attention if nobody saw it.
The method we're building starts from the screens people actually had in front of them and how long they stayed on each. Only then is that time split across what competed on the screen, so an element nobody saw gets no attention, however prominent a prediction says it is.
- Attention counted only where people really looked Building
Every attention and exposure number will link to this method. Accuracy against real eye tracking will be published on How we know once the validation has run.
Ask them why
Interviews that know what each person did.
Behaviour data has no why. Interviews can't see behaviour. RevealWhy is being built to join them: participants try a task on your real site or prototype, with webcam eye tracking if they agree, then a short interview asks about their own moments: what they missed, what they looked at and didn't use.
- Studies with your own customers, through your own link Live
- Real eye tracking in the browser, in studies people join Building
- An interviewer briefed on what each person did and missed Building
- One report joining seen, did and said Building
- Interviews in your participants' language Building
Know it worked
Worked, didn't, or can't tell yet.
The check we're building compares a change with your other pages over the same days, so a sale or a season doesn't pass for a win. Every result will say how sure we are, and when the data can't carry an answer yet, it will say so, and when there will be enough.
- Every change checked against your other pages: worked, didn't, or can't tell yet Building
- No inflated wins Building
- Money only from confirmed orders Building
- Evidence on every answer, and "can't tell yet" when we can't Building
Hand it off
Your coding agent gets the fix. The check runs itself.
The hand-off we're building sends findings to Claude Code or your own agent through MCP, or to your developer or agency, and starts the check on its own when the fix ships. In the pilot, agents also try your key journeys on each release before your customers do.
- Your coding agent gets the fixes Building
- The check runs itself after a fix Building
- Set it up from Claude Code Building
- Send a finding to Slack, Jira, Asana or Notion Building
- Release check: key journeys tried on every release, stopping before payment Pilot
Ask your first question.
One question, one page, and the evidence for the answer.
Sign-up opens shortly.