Reversal check: is the change real or the mix?
The overall bounce rate can rise while it falls in every group, because more visits came from a group that always bounces more.
Whether an overall change in bounce rate reverses inside every group, Simpson’s paradox, so you know whether to fix pages or look at the mix.
Why quietstats: you get this without cookies, a consent banner or storing IP addresses, in every plan including Free.
In every plan, including Free. Data in and out.
📘 Step-by-step guide, with screenshots and diagrams →
Start free, no cardSee live demo
- No cookies or consent banner
- No IP addresses stored
- Groups under 5 visitors hidden
- 845-byte script
- Free plan, no card


What it is
quietstats compares the overall bounce rate with the period before, and for each split checks the typical change inside groups with 10 or more visits in both periods. If the overall change goes one way and the groups go the other, it is a reversal.
Why it matters
The overall bounce rate can rise while it falls in every group, because more visits came from a group that always bounces more. A reversal means: do not fix the pages, look at the mix.
How it works
- Open your dashboard, go to Deep dives (or Reports on a phone) and open Analysis lab.
- Under Tools, tap Reversal check. You can also search for it on the All reports screen.
- Pick a date range (and any options at the top), read the “How to read this” notes, and download any table as CSV.
Privacy: what is and isn’t collected
Collected
- Nothing new: it reads the pageviews, events, goals or daily totals you already have
Never collected
- Anything about a single visitor
- Links between days: visitor ids still reset at midnight
- Cookies, localStorage or anything else stored on the visitor’s device
- IP addresses or user agents (used in memory, then discarded)
- Anything that links a visitor across days (the visitor ID resets daily)
Rows and limits
- Needs data from the period before your range
- Groups need 10 or more visits in both periods
- Up to 92 days and 200,000 pageviews per screen
- Read from the database only when you open it, then kept for 10 minutes
“Rows” are database rows written, which is what our hosting is billed on. Reads are not rows written. Methodology →
Questions
How is this different from Mix effect?
Mix effect splits a change into mix and rate parts. This checks for the extreme case where the two point in opposite directions.
Does it store anything new?
No. It is worked out from the pageviews and events you already have when you open the screen, and only the totals are kept for 10 minutes.
Related features
See your trafficWhat changed, linked to traffic
See your trafficCompare any two date ranges
See your trafficHide-list suggestions: referrers that look like spam
Data in and outAre these linked? Splits that go together
Data in and outFriendly name suggestions: readable names for your busiest addresses
Data in and outLook-alike weeks: past weeks shaped like this one
Data in and outClick-to-visit check: how many paid clicks arrive
Sources and campaigns
All data in and out features → · All features → · Docs: Reversal check guide → · Compare features with other tools →
Guides on this topic
Set it up
Nothing to set up: it works with the standard one-line script.
<script data-host="https://quietstats.co.uk" data-dnt="true" src="https://quietstats.co.uk/js/script.js" id="YOUR_SITE_ID" async defer></script>Try it on your own site
Free during launch: one site and 5,000 pageviews a month, no card needed. After launch, paid plans start at £5 a month.