# Heatmaps per variant: where attention actually went > One heatmap of your product page tells you people click buttons and scroll. Two of the same page, one per variant, tell you where your change moved attention and what it pushed out of sight. - Source: https://bixlow.com/en/blog/heatmaps-per-variant - Language: en - Published: 2026-10-26T09:00:00.000Z - Updated: 2026-10-06T09:28:18.400Z - Author: fernando - Topics: session recordings, Microsoft Clarity, mobile CRO, A/B testing, experiment design, CRO, audience targeting - Publisher: Bixlow --- People who take measurement seriously tend to look down on heatmaps, and mostly with reason. A heatmap of your product page shows that people click the buttons and scroll down, which you already knew. Comparison is what makes them useful. Put two heatmaps of the same page side by side, one per variant, same period, same device, and a decorative image becomes a measurement of where your change moved people's attention. ## Three maps, three questions Scroll depth shows what share of visitors reached each point down the page. It's the map that most often explains a null result. If your change sits at 70% of the page and only 30% of visitors get that far, most of your sample never saw what you changed. Comparing the scroll curves of the two variants also shows whether the change made people explore further or give up sooner. Click maps show where people tap or click. Check first whether your new element gets any interaction at all. Then look for something more useful: clicks on things that aren't clickable. Images people expect to zoom, headings they expect to expand, prices they expect to open a size chart. Those dead clicks are direct evidence of an expectation the page isn't meeting. Attention maps show where people paused. They're mostly good for checking whether a block you added is being read or skipped, and whether it took attention away from something that used to get it. ## Always read them in pairs "Only 22% of people reach the reviews" means very little on its own; it could be perfectly normal for that page, that device and that category. "22% in variant B against 41% in control" tells you your change put the reviews out of reach for half the people who used to see them. Reading heatmaps as differences is the one habit that makes them worth opening, and it's ignored all the time. In Bixlow, each result links to Clarity already filtered to each variant, so the side-by-side takes a couple of clicks instead of becoming a data project. ## Narrow the slice first A heatmap that mixes mobile and desktop is no use at all. The layouts are different, and averaging the two curves produces a shape that describes neither. New and returning visitors are similar: returning visitors scroll faster and further because they know where things are, so mixing them flattens precisely the difference you're looking for. Filter to one variant, one device and one visitor type, then compare with the same slice of control. Anything broader and you're looking at an average of groups that behave differently. ## What to check after a test Start with whether the change was seen at all. Look at scroll depth at the position of your change in both variants. If only a third of people got there, two-thirds of your sample diluted the result, and the real effect on the people who saw it is much bigger than the headline figure. Then check whether it pushed something else down. Adding a block moves everything below it, so compare scroll depth at the old elements. A trust block that builds trust and drops the add-to-cart below the fold can easily come out negative overall. Look for new dead clicks. If your variant added something that looks interactive and isn't, the click map shows it straight away, and it's often a five-minute fix that turns a losing variant into a winning one. And look for rage clicks, which Clarity flags. A cluster that shows up in one variant and not in the other is a bug, whatever it looks like in the numbers. ## Where heatmaps stop Heatmaps aggregate, so they hide the individual path. They'll show you that a region drew attention but not in what order, or what the person was trying to do. For that you need to watch sessions. Nor do they prove an effect. A variant whose heatmap looks better on every count can still lose the test. Let the experiment make the call and use the heatmap to understand it. --- ## Meet our AI CRO agent What you just read, put to work by an agent: it understands your visitors, decides what to test or personalize, builds it on your site without a developer and tells you when a result is real. https://bixlow.com/en/ai-cro-agent