Heatmap Reading and Heuristic Review of Revenue Templates

Scroll and click maps are evidence about placement, and nothing else. Most of the heatmap findings clients hand me turn out to be artefacts: a click map aggregating a sticky element across four page templates, a scroll map computed on sessions that never rendered because of a lazy-load race, an attention map read across two Android viewports whose fold sits 340 pixels apart. This review separates artefact from finding first, then reads what survives against a heuristic model of the page. The heuristic pass: I work through the templates carrying your revenue — category, product, cart, and for an app, the store listing and the first-run screens — against a checklist written for the vertical rather than a ten-point usability list that fits any site. For furniture that means dimension legibility, where the delivery promise sits relative to price, whether material and swatch information survives the jump to mobile, and whether the assembly commitment is discoverable before the add to cart or after it. For pet products it means the frequency selector, weight and breed sizing, and how visible cancellation is at the moment someone commits to autoship. Every observation is written as a claim that could be wrong, with the thing that would falsify it. The tools I read: Hotjar, Microsoft Clarity, Contentsquare and FullStory, plus whatever you already run. Clarity's rage-click and dead-click flags are a search index, not findings; they point at candidate pages and each one gets verified by hand, because dead-click detection fires constantly on non-interactive elements people tapped for reasons that have nothing to do with a defect. Maps get segmented by device class and by new against returning before anyone reads them. An aggregate map across both is close to meaningless once returning customers pass about 30 percent of sessions, and on a repeat-purchase catalogue they always do. The document: Page by page, annotated, screenshots with the specific region marked. Each issue is labelled one of three ways: verified in behavioural data, inferred from the heuristic model, or unresolved and needing session review. The third label appears often and that is deliberate — anything I cannot support, I say I cannot support. Issues are then grouped by whether they are a fix, a test, or a question, and the fixes carry enough detail for a front-end developer to pick up without booking a second meeting. Not part of this: No test design, no statistics, no implementation. No opinion on your brand or visual identity. I will not rewrite the page and I will not send a mockup. A heuristic review whose output is a new design is a redesign sold with extra steps, and once the deliverable is a mockup nobody ever checks whether the diagnosis was right. Wrong fit: Pages under about 3,000 sessions in the sampling window. The maps will be noise and you should be booking moderated sessions instead. Teams who want a scored report with a number out of 100; I do not produce those, because the weighting behind the score is invented and the number pulls every conversation away from the specific broken thing. And anyone who wants this to settle an internal argument about a hero image. I will tell you what the evidence supports, which is usually less than either side of that argument wants, and on hero images it is frequently that the element sits below the attention threshold of most of your mobile traffic and the argument was not worth the meeting.