▤The Test Group experiment running Open the partner account
paid
Affiliate disclosure. The partner link in the masthead and in the band beside the copy on this page is a sponsored link to a partner operator, and this site may be paid if you open an account through it, at no extra cost to you. It carries rel="sponsored noopener" and opens in a new tab. A desk about how a screen is measured and tested should not leave its own funding unsaid: one link funds the site, no operator and no product is named, rated or recommended anywhere on it, and this site runs no analytics of its own on its readers.
The Test Group / The replay
When the counts become a picture of one person’s screen

The session replay, and who may watch it

A replay is the same events, drawn. Instead of a count of clicks it is a film of one screen for one session, with the mouse path, the pauses and the fields as they were filled. It is the most useful instrument in the loop and the most invasive, and on the samples almost none of it is ever watched.

Desk spec
recordings a week
10,000
each
3.4 MB
retention
30 days
watched
2.1%
the eventOne interaction, written down. A click carries a name, a time, an account, a session and what was on the screen, and it is kept whether or not the reader chose to be measured.
the funnelThe order the steps happen in. A landing visit becomes an account, an account becomes a deposit page, and 100,000 visits end in 2,074 first bets - a 2.1% path the whole loop is aimed at.
the armsTwo versions of one screen shown at the same time, and a rate for each. 4.30% against 5.16% is a 0.86-point lift, and the interval decides whether it is a result or a coincidence.
Direct answer

A session replay is a recording of one reader's screen, rebuilt from the event stream rather than filmed: the pointer path, the scrolls, the clicks, the pauses and the form fields as they were typed. On the samples 10,000 recordings a week weigh 34.0 GB, a 30-day retention holds 145.9 GB, and 2.1% of them are ever watched.

How a replay is made without a camera

There is no video. The page announces its own changes to a script, which writes them down as events, and a player redraws the session later from that list. That is why a replay is small, why it can be searched by event, and why it can be replayed for a screen the operator no longer ships.

Sample D - one week of recordings, and what each part costs
ItemValueNote
recordings a week10,000a sample of sessions, not all of them
size per recording3.4 MBa 14-minute session at 8 frames a second
a week of storage34.0 GB10,000 x 3.4 MB
retention30 days4.29 weeks of recordings held at once
held at peak145.9 GB34.0 x 4.29
recordings ever watched2102.1% of the week
mean share of a session watched12.9%1.8 of the 14 minutes
sample D - the storage and the attention a week of recordings = 10,000 x 3.4 MB = 34,000 MB = 34.0 GB retention = 30 days = 4.29 weeks held at peak = 34.0 x 4.29 = 145.9 GB watched at least once = 210 / 10,000 = 2.1% mean watched per replay= 1.8 of 14 minutes = 12.9% so of 140,000 minutes recorded in a week, 210 x 1.8 = 378 minutes are ever looked at, which is 0.27% of what was stored the recording is kept in full and read almost not at all.

What is in the recording, and what is taken out

A replay is only useful if it shows the fields, and it is only tolerable if it does not show them. That tension is resolved by masking, which is a decision made field by field: on the samples three of the four completed fields are masked and the fourth is not.

Sample D - the four fields of one session, as they appear in a replay
FieldIn the replayWhy
a card numbermasked, all but the last foura payment credential, kept out by rule
a date of birthmasked entirelyan identity field, needed once and not twice
a deposit amountmasked in the replayread from the count instead
an e-mail addressshownthe sample's weakest decision, and the one a complainant would find first
four fieldsthree masked, one notmasking is a design choice, not a property of the tool
sample D - what the unmasked field is worth sessions with a completed e-mail field = 1,800 of 10,000 = 18.0% of those, the field appears unmasked in the replay replays watched at all = 2.1% replays watched that show the field = 0.021 x 0.18 = 0.38% so 38 of the 10,000 recordings expose an address to anybody who watches, and the retention window is what limits it afterwards: at 30 days, the exposure ends with the recording, unless the excerpt was exported before it expired.
The numbers are invented and the masking decisions are the samples'. Whether a named operator records sessions, what it masks and how long it keeps the recordings are answered only in its own documents, and a reader who wants the answer should ask it in writing rather than infer it from a page like this.
Finding out whether a site records your session
  • Look for the phrase "session replay" or "session recording" in the privacy notice, not just "analytics".
  • Check the retention line for recordings separately from the one for counts.
  • Check whether the consent you gave covers recording, because the two are often bundled.
  • Ask whether an excerpt can be exported from the tool, since that outlives the retention window.
  • Use the data-access route and ask specifically for any recording, which is a request few readers make.

Read next

The eventsThe stream a replay is rebuilt fromThe consent lineWhat may be recorded at allThe vendorsWhere the recording is heldThe recordWhat is never published back