Session replay

The page as rendered, including content that is specific to a signed-in visitor.

Pointer movement, scroll position and click targets, which together reconstruct attention as well as action.

Keystrokes in form fields, unless masking is configured — and masking is configuration, not a default in every product.

The evolution of the page over time, which means content that appeared briefly and was dismissed is still in the recording.

Every payment field, which in most products is masked by default and should be verified rather than assumed.

Authentication fields, and any field where someone might paste something they should not — a password field is obvious, a search box is not.

Content that identifies a person: an account page, an order history, a message thread.

Anything a third party told you in confidence, which on a support interface is most of the screen.

Masking is per-selector work, which means a new page or a redesign can quietly un-mask something that was covered.

It processes far more than a measurement cookie: the content of a session rather than the fact of a visit.

It is a third-party script by construction, sending that content to another company's infrastructure.

It belongs in the privacy documentation explicitly and by name, rather than folded into a general reference to analytics.

None of which makes it illegitimate — it answers questions no other tool can. It makes it a decision that deserves the deliberation it rarely gets.

The presence of a session replay product, detected from real script hosts and markup fingerprints rather than inferred.

Whether it loads before a consent choice has been made.

Whether a cookie or privacy policy is published that could describe it.

Not what is masked, not what is recorded, and not what is retained — all of which are configuration inside the product.

Diagnosing a specific reported problem: someone says a form would not submit, and the recording shows what they actually did.

Finding where an unfamiliar interface confuses people, which aggregate analytics can only hint at.

Both are narrow uses, and both are better served by recording a sample for a limited period than by recording everything indefinitely.

Aggregate interaction measurement — scroll depth, rage clicks, funnel drop-off — answers most of the same questions without capturing content.