Email Heatmaps: Understanding Subscriber Behavior
What click maps show, what they cannot show, and how to read one without over-fitting to forty clicks.
On this page
The short version
- An email heatmap is a click map, not an attention map. It cannot show what was read, only what was clicked — which is a much smaller thing.
- Below a few hundred clicks the map is mostly random. Patterns appear in noise readily, and a map with forty clicks on it will show you something whatever the truth is.
- The most useful reading is comparative: the same template across many sends, looking at which zones consistently earn clicks.
Click maps are the most over-interpreted artefact in email reporting. They look like the heatmaps used in web analytics, which track cursor movement and scroll depth, and they are not that — an email cannot observe anything except the links that were followed.
That makes them genuinely useful for a narrow question and misleading for the broad one people usually ask them.
What the map contains
One data point per click: which link, and how many times. Everything else — the colours, the gradients, the sense of a region being warm — is a rendering of that, and there is nothing underneath it about reading, hovering, or attention.
How many clicks before it means anything
The rule of thumb worth holding is that a difference between two zones is only interesting if it would survive a plausible amount of random variation. With forty clicks spread across six links, almost any arrangement is plausible.
A few hundred clicks on a single send is where patterns start to be worth reading, and even then only for large differences — one zone getting three times another, not one getting fifteen percent more.
Below that, the honest use is to aggregate several sends of the same template rather than reading each one.
A typical distribution across a template's zones
Aggregated across many sends of one template, this is the shape that usually appears — and the last slice is the one that surprises people.
Illustrative aggregate shape for a single-offer template, not measured data. Build your own from six sends of the same template before drawing conclusions.
The questions a click map can answer
Is the primary action getting the majority of clicks, or is it competing with something? Does the closing line earn enough to justify keeping it? Are readers clicking a link that was meant to be incidental? Is one zone consistently dead across many sends?
Each of those is a structural question about the template rather than a question about one send, which is why the aggregate view is the useful one.
The questions it cannot answer
Whether anyone read the middle section. Where attention went. Why a zone got no clicks — it might have been unread, or read and unpersuasive, or read and fully satisfying, and the map treats all three identically.
That last case matters. A section that answers a reader's question completely produces no click, and looks exactly like a section nobody read.
What to do with a dead zone
The instinct on finding a zone with no clicks is to remove it. That is right about a third of the time and wrong the rest, because three quite different situations produce an identical empty region on the map.
It may contain no link at all, in which case the map is reporting a fact about your markup rather than about your readers. It may be answering the reader's question so completely that no click is needed — a shipping cost stated plainly produces no click and prevents a support enquiry. Or it may genuinely be unread.
Distinguishing them takes one step: check whether removing it changes the click rate of the zone beneath it. If the primary button improves when the dead section goes, the section was in the way. If nothing moves, it was doing no harm and possibly some good.
Reading a click map honestly
- At least a few hundred clicks, or aggregated across sends
- Navigational and footer clicks separated from the primary action
- Scanner clicks filtered on business lists
- Comparisons drawn between zones, not between individual links
- No conclusion drawn about what was read
- Dead zones checked across several sends before removing anything
Frequently asked questions
Can I get scroll depth for an email?
Not reliably. Some techniques infer it from image loading order, and privacy pre-fetching has made those largely useless. Treat any tool promising email scroll data with scepticism.
A section gets no clicks. Should I cut it?
Not on that alone. Check whether it contains a link at all, whether it might be answering the question rather than prompting one, and whether the pattern holds across several sends.
Do click maps work with mirrored links?
Only if your platform tracks each instance separately. Many collapse identical URLs into one entry, which destroys exactly the placement information you were looking for — check before designing a template around it.