What a heatmap actually shows — and what it doesn't
You installed a heatmap tool, waited a few days, and opened the report: one button is glowing red, the block underneath is solid blue. The conclusion looks obvious — red means people love it, blue means they ignore it, so the blue block should go. A few days later you've redesigned the page, and the sales number hasn't moved at all.
The problem isn't the map. It's how it gets read. A heatmap is a genuinely useful tool, but it only answers one question: WHERE was the visitor. WHY they were there, what they felt, why they didn't click — that takes a different kind of evidence entirely. Here's the distinction, worked through step by step: what a heatmap actually measures, when the colour lies to you, and how to turn a red patch into something you can act on.
What a heatmap actually measures
"Heat" doesn't describe one visitor's behaviour — it describes thousands of visits stacked on top of each other. The tool logs the position of every click and every scroll, then compresses all of it into a single image: the warmer the colour (the closer to red), the more activity happened at that spot. That compression comes in three main forms.
Click maps — who taps where
A click map shows how many times each element got tapped, and not just buttons and links — per Microsoft Clarity's own documentation, it also catches elements users mistook for a link that aren't one. That's the fastest way to spot where an interface is misleading someone. The same documentation splits clicks into subtypes: a "rage click" — a visitor tapping the same spot repeatedly in a short window, usually because something isn't responding; a "dead click" — a tap that produces no reaction at all; an "error click" — one immediately followed by a JavaScript error. All three can look like the same shade of red on screen, but they mean opposite things — one of them is very much not satisfaction.
Scroll maps — how far people actually get
A scroll map shows what percentage of visitors reached which part of the page. The useful number here is the "average fold" — the line most visitors see before they start scrolling. If your main offer sits below that line, most of the audience never lays eyes on it, no matter how well it's written.
Attention and area maps — where the time goes
An attention map, unlike a click map, shows how much TIME visitors spend looking at a given part of the screen — and the most-looked-at zone doesn't always match the most-clicked one. An area map adds up every click inside a block you select — a whole navigation bar, say — rather than element by element, which is useful when what matters is the block's overall weight, not any single item in it.
The colour alone tells you nothing — context does
A red patch doesn't mean "people love this," it means "a lot happened here." The rage-click example above proves that in the cleanest possible way: same red, opposite meaning. By the same logic, the first field on any form is almost always the "hottest" one — not because it's especially interesting, but because it's where everyone starts filling it in. A nav menu runs hot for the same reason: it's on every page and picks up repeat clicks by default, which says nothing about how good the menu itself is. The first question when a heatmap opens should be "why did this get hot," not "this is hot, so it must be working."
Why it doesn't answer "why"
A heatmap compresses thousands of visits into one picture — that compression is exactly what makes it useful, and it's also its limit. Compressing removes individual intent: out of a hundred people clicking the same spot, one was genuinely interested, one misclicked, one was hunting for the price, one was just testing whether the screen responds. On the map, all of them show up as the same red pixel.
Ignoring that gap costs real time and money. A block gets labelled "not working" and cut, when the actual problem sat one step earlier — in whatever sent the visitor there in the first place. The redesign takes two or three weeks to build, and the number doesn't move, because the real cause was never touched.
A heatmap answers WHERE — it does not answer WHY. Finding the why is a separate, necessary step.
Turning a red patch into a hypothesis
A heatmap is not a decision by itself — it's the starting point for a question. These four steps turn it into something worth shipping.
1. Write the question before you open the map
Before you look, put in one sentence what you actually want to know about the page — "do visitors reach the pricing block," "is the main button getting noticed." Without a question, your eye locks onto the reddest spot and calls that the answer.
2. Filter by segment
Blending every visitor into one map produces a distorted read. Mobile and desktop need separate maps — screen size changes tap position entirely. New and returning visitors also behave differently; splitting them shows which audience is actually getting stuck where.
3. Verify with session recordings
Once you've found a red spot, watch a handful of real session recordings from that exact location. Watching the behaviour instead of a number usually clears up most of the "why" on its own — a visitor hovers over a button, pauses, then leaves because the price text next to it isn't legible.
4. Test with one small change
Once you have a hypothesis, change that one detail — the button's position, its copy, its colour — not the whole page. Run an A/B test if traffic allows it, or a before/after comparison if it doesn't. Even a change that shows no difference isn't wasted effort — knowing which hypothesis was wrong narrows what to try next.
The most common mistake: rebuilding the whole page off one map
Redesigning an entire page's structure off a single heatmap snapshot is the most expensive mistake available — the result doesn't show for weeks, comparing it against past data gets harder, and the real cause is usually left untouched. A heatmap is a research stage, not a final decision; confirming each loss point with a small, measurable step is cheaper and faster than changing everything at once.
If you want to read a heatmap yourself, tools like Microsoft Clarity are — per their own documentation — free forever, and take a few minutes to install. Turning the red-and-blue picture into a ranked, concrete list of fixes is a separate job, though: which loss point to fix first, which hypothesis a session recording has already confirmed, and which change actually needs a test. The team that runs that sequence for you masks sensitive fields, tracks until enough data has built up, and hands back the findings as a prioritised list.
