Journal · 18 August 2025
Session replay is not analytics
A video of a thumb is vivid. Vivid is not the same as true enough to brief an engineer.
I have watched teams paste a rage-click clip into Slack and call it a trend. The clip was one person, on one OS, after a font size change they had not consented to in a lab. Session replay is a lamp. App Analytics is a census with declared absences.
Start with a question and a sample rule
Before anyone presses play: what would change if we saw this ten times, or a hundred? How are sessions chosen — random among a funnel step, or the vendor’s “frustrated” score? Frustration scores are productised adjectives. Write your own inclusion rule or admit you are browsing.
Masking and lawful basis are not optional colour in the United Kingdom. If you cannot explain to a colleague why a session was recorded, do not record it. Teaching this without a legal opinion is a limit of our workshop; we will say so in the room.
Where replay earns its keep
After a defined drop in a named event, a small sample can show whether the copy, the keyboard, or a webview is the culprit. That is illustration. The count still lives in the stream. Our Session forensics clinic is two days of insisting on that order. If you need the calendar, the programmes list has the informational fee.