SEO Analytics & Reporting
Universal Analytics vs GA4: What Actually Changed
If your GA4 numbers look wrong next to the old Universal Analytics reports, they’re not. The definitions changed, not your traffic. Here’s exactly what changed, from the event model to engagement rate replacing bounce rate, so you can read GA4 without dragging the wrong instincts across from UA.
By Rahul Saini, Author at Search Counsel Co. Last updated [July] 2026.
Featured answer: what changed from UA to GA4?
GA4 changed the definitions, not just the interface. Everything is now an event instead of a session or pageview, bounce rate was replaced by engagement rate on entirely different rules, and sessions and users are counted differently. That’s why your GA4 numbers look off next to old Universal Analytics reports: the users didn’t change, the measurement did.
Don’t compare GA4 and UA numbers like-for-like. Your figures differ because the definitions changed, not because your traffic did, and Google itself says to expect discrepancies of up to around 20% in users and sessions. Putting GA4 numbers in a continuity chart next to old UA numbers, as if they mean the same thing, is the most common reporting mistake of the whole migration.
Model
Events, not hits
Everything is an event now, not a session or pageview.
Quality
No more bounce
Engagement rate replaced bounce rate, on new rules.
Sessions
Counted anew
No midnight or source resets, so counts shift.
Users
Active by default
GA4 shows active users, so totals look lower.
What’s in this guide
The translation table
If you knew UA, here’s how the familiar metrics map onto GA4, and why each one reads differently:
| Universal Analytics | GA4 equivalent | Why it differs |
|---|---|---|
| Bounce rate | Engagement rate (bounce = its inverse) | Redefined entirely; not comparable to UA bounce |
| Sessions | Sessions | No longer reset at midnight or on source change; usually fewer |
| Users | Active users (Total users = old Users) | Active users run lower than UA users |
| Goals | Key events | Renamed and rebuilt on the event model |
| Pageviews | Views | Now just another event |
| Exit rate | Not in default reports | Approximate via Path exploration |
From hits to events
This is the root of everything else. Universal Analytics was built on hits, pageviews, events, and transactions, with the session as the central unit. GA4 is fully event-driven: every interaction is an event with parameters, and even a pageview is just an event called page_view. That shift is what lets GA4 track across devices and suits a privacy-first era, but it also means the old report structure and metric definitions don’t carry over. If a number looks strange, the model change is usually why.
Bounce rate to engagement rate
This is the change that confuses the most people. In UA, a bounce was any single-page session with no further interaction, so a visitor who read a 12-minute article and left counted as a bounce. That made bounce rate almost useless for content sites.
GA4 replaced it with engagement rate, built on the idea of an engaged session:
A session is “engaged” if any one of these is true:
- It lasted longer than 10 seconds (the default, adjustable up to 60), or
- It included at least one key event (a conversion), or
- It had two or more pageviews or screenviews.
Engagement rate is engaged sessions divided by total sessions. GA4’s bounce rate is simply what’s left over: 100 minus engagement rate.
Here’s the trap that catches almost everyone: GA4 engagement rate is not the inverse of UA bounce rate. They measure different things. UA bounced a one-pageview session no matter what; GA4 counts that same session as engaged if it lasted ten seconds. So you cannot recover your old UA bounce rate by subtracting engagement rate from 100, and you shouldn’t try. Because the bar is more lenient, your GA4 bounce rate looks far lower than UA’s, an 80% UA bounce can read as 40 to 50% in GA4. Don’t celebrate a fixed bounce rate; only the definition changed.
One honest caveat: engagement rate depends on choices you make. Because a key event marks a session engaged, what you flag as a key event directly moves the number. Mark something trivial like a shallow scroll as a conversion and your engagement rate inflates. Read it alongside engagement time and key events, and against the Landing page dimension rather than individual page paths, or it misleads.
Sessions and users, redefined
Two more quiet changes explain most remaining discrepancies.
Sessions no longer reset at midnight or when the traffic source changes, as they did in UA. One GA4 session can now span sources, which is why session counts differ and why you must pair the sessions metric with the session-scoped Session source and Session medium dimensions, not the event-scoped ones, or the numbers won’t line up.
Users now default to active users, people with at least one engaged session, in most reports. UA’s “Users” is closer to GA4’s “Total users.” So GA4’s headline user count runs lower than UA’s, which is normal and not lost traffic. Between these two changes, GA4 typically reports somewhat fewer sessions and users than UA did for the same site.
What to do about it
- Stop comparing periods across the switch. Don’t chart GA4 against UA as one continuous line. Annotate the changeover and treat the two as different rulers.
- Rebaseline your benchmarks. Set fresh targets on GA4 numbers rather than UA-era ones, especially for engagement and bounce.
- Export your UA history before it’s gone. Universal Analytics data is being permanently deleted, so if you still need it, get it into BigQuery before the deletion date rather than after.
- Use engagement rate as your quality signal, read with context, and lean on the five metrics that matter for SEO rather than chasing old UA figures.
FAQ
Is GA4 bounce rate the same as Universal Analytics bounce rate?
No. UA bounce rate counted any single-page session as a bounce; GA4 bounce rate is just 100 minus engagement rate, and a session counts as engaged after ten seconds, one key event, or two pageviews. The two are different metrics, so GA4 bounce rates read much lower and can’t be compared to UA’s directly.
What is a good engagement rate in GA4?
It varies by site and page intent, so there’s no universal target. Content sites with many single-page visits look different from ecommerce sites where people move between pages. Read engagement rate alongside engagement time and key events, and compare a page to its own trend and its siblings rather than to a fixed benchmark.
Why does GA4 show fewer sessions and users than UA did?
Because both are counted differently. GA4 sessions no longer reset at midnight or on a source change, and GA4’s default user metric is active users rather than total users. Google says to expect discrepancies of up to around 20%. It’s a measurement change, not a drop in traffic.
Can I still see bounce rate and exit rate in GA4?
Bounce rate exists but is hidden by default; you can add it to a report or exploration, though it’s just the inverse of engagement rate now. Exit rate was removed from standard reports and can only be approximated, for example through Path exploration. Google wants engagement rate to be the primary quality signal.
Do I still need my old Universal Analytics data?
If you want year-over-year history that predates GA4, yes, and it’s being permanently deleted, so export it to BigQuery before the deadline if you haven’t. Just remember you can’t compare it like-for-like with GA4 because of the definitional changes covered above.
The one thing to do next
Stop reading GA4 through UA eyes. Rebaseline your engagement and bounce expectations on GA4’s own numbers, annotate the switch in any long-term chart, and if you haven’t exported your UA history, do it before it’s deleted. Then focus on the metrics that actually matter for SEO reporting.
Sources used
Metric definitions and the UA-to-GA4 differences follow Google Analytics documentation, including Google’s own guidance to expect up to roughly 20% discrepancy in user and session counts. The engaged-session criteria and the bounce-versus-engagement distinction are from Google’s Help documentation and industry analysis. The Universal Analytics data-deletion timing should be re-verified against Google’s current notices before publishing, since the date is fixed but worth confirming.