GA4 vs Universal Analytics: the differences that matter
Tomorrow, 1 July 2023, standard Universal Analytics properties stop processing new data. From then on, GA4 is Google Analytics.
If you’ve already moved, you’ve probably had the moment: you open GA4, compare a number with what UA said for the same week, and they don’t match. Users are different. Sessions are different. Bounce rate either vanished or means something else. It’s easy to assume something is broken.
Usually nothing is broken. GA4 is a different tool that counts things differently. These are the differences that change how you read your numbers.
1. Everything is an event
UA was built around sessions and pageviews. It had different “hit types”: pageviews, events, transactions, social interactions. Events had a category, action and label.
GA4 is built around events. A page view is an event (page_view). A scroll is an event. A click on an outbound link is an event. Each event can carry extra details called parameters.
Why it matters: the reports you relied on in UA were built around sessions. In GA4, the data model is more flexible, but you often have to decide what you want to track and how. I covered the first events to set up in GA4 events for beginners.
2. Engagement instead of bounce rate
In UA, a bounce was a single-page session with no other interaction. Someone could read your whole article for ten minutes and still count as a bounce.
GA4 uses engaged sessions: a session that lasted longer than 10 seconds, had a conversion event, or had two or more page views. Engagement rate is the share of sessions that were engaged. GA4’s bounce rate is now simply the opposite of that.
Why it matters: GA4’s bounce rate is not comparable with UA’s. Many content sites will see a much “better” figure in GA4, and that’s because the definition changed, not the content.
3. Users and sessions are counted differently
GA4 focuses on active users in many reports, and it counts sessions differently from UA (for example, a session doesn’t restart just because the date changes at midnight or someone arrives from a new campaign).
Why it matters: expect user and session counts to differ from UA, sometimes noticeably. Don’t compare a GA4 month with a UA month and draw conclusions. Compare GA4 with GA4.
4. No views, just data streams
UA had properties and views. Many sites used views for filtered data: one view excluding office traffic, one only for a subdomain, a raw backup view.
GA4 has properties and data streams (web, iOS, Android). There are no views. Some filtering exists (for example, excluding internal traffic and developer traffic), but much of what views did now happens in reports and comparisons instead.
5. Goals are now conversions
In UA you set up goals: a destination page, a duration, an event. In GA4, you mark an event as a conversion. If you want a “thank you page” goal, you create or choose an event for it, then mark it as a conversion.
Why it matters: your UA goals did not move across automatically unless you migrated them. Check what’s marked as a conversion in GA4 now, because it’s what most reports use.
6. Data retention is short by default
This one catches people out. GA4 keeps user-level and event-level data used in Explorations for 2 months by default. You can change it to 14 months under Admin → Data Settings → Data Retention.
Standard reports aren’t affected in the same way, but if you want to build detailed explorations over a year, change this setting today. It isn’t backdated.
7. Reports are leaner; Explorations do the heavy lifting
GA4’s standard reports are simpler than UA’s long menu of reports. For deeper analysis, there’s Explore: free-form tables, funnels, path analysis and segment overlap.
There’s also a free BigQuery export for standard GA4 properties, which used to be a paid (360-only) feature in UA. Most small sites won’t need it, but it’s there.
Why it matters: if you can’t find “your” report, it might not exist as a default. You can customise the report library, or build it in Explore or Looker Studio.
8. Privacy and thresholds
GA4 was designed with privacy rules and cookie consent in mind. It doesn’t log or store IP addresses, and in some reports you’ll see data hidden because of thresholds when numbers are small or Google signals are on.
Why it matters: small sites sometimes see “missing” data in demographic or detailed reports. That’s often thresholding, not a tracking fault.
What I’d do this week
- Set data retention to 14 months if you haven’t.
- Check your conversions in GA4 match what you tracked as goals in UA.
- Export the UA reports you rely on before access ends. Google has said you’ll keep access to previously processed UA data for at least six months, but don’t leave it to the last day. I’ll go through how in a follow-up post.
- Write down the new definitions for anyone who reads your reports, so nobody panics when “bounce rate” halves.
- Start fresh comparisons. From July, compare GA4 with GA4.
Related reading
- Universal Analytics is ending: start your GA4 migration now
- GA4 events for beginners: what to track first
- The GA4 migration checklist to finish before 1 July
- Measuring content performance without vanity metrics
Which UA report will you miss most? Find its closest GA4 version today, while you can still open UA and compare the two side by side.