Session replay and heatmaps

Session replay records how individual visitors interacted with a page so the session can be watched back, and heatmaps aggregate those interactions into a visual map of where visitors clicked, moved and scrolled.

Session replay records how individual visitors interacted with a page, capturing mouse movement, clicks, scrolling, form interaction and page changes so a session can be watched back afterwards. Heatmaps aggregate the same interactions across many visitors into a visual map showing where people clicked, how far they scrolled and where attention concentrated.

How does session replay work?

It reconstructs the session rather than filming it.

A JavaScript tag records the page structure and then a stream of events: pointer positions, clicks, key presses, scroll offsets, and changes to the page itself. Playback rebuilds the page from that data and replays the events against it. This is why a replay can look slightly different from what the visitor saw, and why replays of highly dynamic pages sometimes break: the tool is redrawing the page from a description, not playing a recording.

Heatmaps use the same event stream, aggregated. A click map shows where people clicked, including clicks on things that are not clickable, which is one of the most useful signals the format produces. A scroll map shows the share of visitors who reached each depth. Move maps show cursor position, and are the weakest of the three because cursor position is a loose proxy for attention.

What are session replay and heatmaps genuinely good at?

Three things, and they do them better than anything else.

Seeing the unexpected. Nobody hypothesizes that visitors are clicking a static image because it looks like a button. You find it by watching. Aggregate analytics cannot show you a behavior you did not think to instrument.

Confirming a specific bug. When a funnel step drops on one browser and nowhere else, ten replays will usually show you the broken element in a few minutes, which is far faster than reproducing it from a description.

Making a problem undeniable inside the organization. A chart showing form abandonment gets argued with. Watching five people in a row fail at the same field ends the argument. This is a political function and it is a legitimate one.

Who sells session replay and heatmaps?

Hotjar and FullStory are the reference vendors, alongside Microsoft Clarity, Contentsquare, Quantum Metric, Glassbox, Mouseflow and LogRocket. Several analytics platforms now bundle replay as a feature rather than a product.

They differ mainly on scale and rigor rather than on the core idea. The lighter tools are quick to install and priced for a single team. The enterprise tools add sampling controls, quantified impact analysis, retention and compliance features, and integrations that let you jump from a metric to the sessions behind it.

What are the limits of session replay?

It shows what happened, never why. You can watch someone hesitate for eleven seconds on a shipping field and be no closer to knowing whether the cause was cost, timing, a returns question or a payment concern. Those four causes need four different fixes. The replay is identical in all four cases. Teams routinely watch a session, form a confident interpretation, and ship a fix for a cause the visitor never had.

It does not scale to volume. A busy site produces more sessions in an hour than anyone will watch in a month, so in practice you watch a filtered handful. Filtering is a hypothesis, which means you mostly find what you were already looking for.

It carries real privacy weight. Replay captures whatever is on screen, which on many sites includes personal, health or financial information. Masking is available in every serious tool and has to be configured deliberately, field by field, before launch rather than after an incident. Session recording has drawn regulatory attention in several jurisdictions, and consent handling is a decision for legal rather than for the analytics team.

The watching itself is expensive. The recurring cost of this category is not the license, it is the hours of skilled attention that watching consumes, which is why so many replay tools are enthusiastically bought and quietly abandoned in month four.

How Pulse Insights relates to this

We are not a session replay or heatmap tool, and if that is what you need, buy one.

Specifically, buy session replay when you need to reproduce a bug you cannot describe, when you are auditing a redesign before launch, when you need visual evidence of a usability problem to unblock a decision, or when your question is genuinely "what are people doing here." Those are real jobs and replay tools do them well. We do not record sessions and we do not produce heatmaps, so we are not an alternative for any of them.

What we do is the step that replay leaves open. We monitor behavioral signals to detect that someone is stuck, then ask one short contextual question to establish which cause it is, then deliver a response the client's team approved in advance, in the same session. Where replay gives you an eleven-second pause to interpret, we get the visitor to tell you it was the returns policy, and answer them before they leave.

Many teams run both. Replay is the better tool for finding an unknown problem in a small number of sessions. Asking is the better tool for explaining a known problem across all of them.

Frequently asked questions

What is the difference between session replay and a heatmap?

Session replay reconstructs one visitor's session so you can watch it back. A heatmap aggregates the same interaction data across many visitors into a visual summary of clicks, scroll depth or movement.

Does session replay record everything a visitor types?

It can, which is why masking configuration matters. Serious tools let you exclude fields and elements from capture, and payment and health fields should be excluded before launch rather than after a review.

Can heatmaps tell you why visitors leave?

No. They show where interaction concentrated and where it stopped. The cause of a drop is not observable in the data, and inferring one from a scroll map is how teams end up fixing the wrong thing.

Is session replay compliant with privacy regulation?

It depends entirely on configuration, consent and jurisdiction. The technology can be operated compliantly, but capture is the default in most deployments and compliance is not. Treat it as a legal review, not a settings page.

Do you need session replay if you already ask visitors questions?

They answer different questions. Replay finds problems nobody suspected in a handful of sessions. Asking explains a suspected problem at scale. Neither substitutes for the other.

Browse every term in the Pulse Insights glossary.