Quick answer: In GA4, engagement rate is the percentage of sessions that were "engaged" — lasting 10+ seconds, firing a conversion, or having 2+ pageviews. Bounce rate in GA4 is simply the inverse: 100% − engagement rate. This is completely different from Universal Analytics, where a bounce was any single-page session with no interaction. That's why a page that "bounced" in UA can look "engaged" in GA4 — and why you can't compare the two numbers directly.
This is a spoke of the GA4 vs Universal Analytics guide. For the full list of renamed metrics, see the metrics translation table.
How UA defined bounce rate
In Universal Analytics, a bounce was a session that triggered only a single request — one pageview, no other interaction hits. Bounce rate was the percentage of such sessions. The problem: a visitor could read an entire 2,000-word article for five minutes and still count as a bounce, because they never triggered a second hit. UA bounce rate punished single-page intent (blogs, landing pages) unfairly.
What "engaged session" means in GA4
GA4 flips the framing from "did they leave?" to "did they engage?" A session counts as an engaged session if it meets any one of these:
- Lasted longer than 10 seconds (the default; configurable from 10–60s in Admin), or
- Triggered a conversion / key event, or
- Had 2 or more pageviews / screenviews.
So the reader on your article who stayed five minutes is now correctly counted as engaged.
Engagement rate & bounce rate math
The two metrics are mirror images:
- Engagement rate = Engaged sessions ÷ Total sessions
- Bounce rate (GA4) = 1 − Engagement rate
Example: 1,000 sessions, 700 engaged → 70% engagement rate, 30% bounce rate. They always sum to 100%. Reporting both is redundant — pick one.
Why your bounce rate cratered
If you migrated and your bounce rate dropped from ~60% (UA) to ~20% (GA4), nothing broke. The definitions are different:
| UA bounce | GA4 bounce | |
|---|---|---|
| Trigger | Single hit, no interaction | Not engaged (under 10s, no conversion, <2 views) |
| 5-min single-page read | Counts as a bounce | Counts as engaged (not a bounce) |
| Typical value | Higher | Lower |
The 10-second time threshold alone rescues a huge share of sessions that UA would have called bounces, which is why GA4 bounce rates are structurally lower.
Which metric to actually use
Use engagement rate. It's the native GA4 metric, it's positively framed (higher is better), and it's available across reports and Explorations. Reserve bounce rate for when a stakeholder specifically asks for it — and when you give it, add the caveat that it isn't comparable to the old UA figure.
Two practical tips:
- Adjust the engagement threshold only with intent. Raising it from 10s to, say, 30s will lower your engagement rate across the board — useful if 10s is too generous for your content, but it breaks comparability with past data.
- Don't optimize the metric, optimize the experience. You can trivially inflate engagement rate by firing an early event; that's gaming, not improvement.
If your engagement numbers look implausible, the cause is often a tracking issue (duplicate page_views inflate; broken tags deflate). A quick GA4 audit — or an automated one from Snifflytics — will tell you whether the metric or the implementation is the problem.
FAQ
Does GA4 have bounce rate?
Yes, it was re-added after launch, but it's defined as 1 − engagement rate, which is different from UA's bounce rate.
What counts as an engaged session in GA4?
A session that lasts 10+ seconds, triggers a conversion/key event, or has 2+ pageviews — any one of those qualifies.
Why is my GA4 bounce rate so much lower than UA?
Because GA4 counts a session as engaged after just 10 seconds, so many sessions UA labeled bounces (e.g. a long single-page read) are now engaged.
Should I use bounce rate or engagement rate?
Engagement rate. It's the native, positively-framed GA4 metric. Bounce rate is just its inverse and isn't comparable to UA.
Can I change the engagement time threshold?
Yes — in Admin you can set it from 10 to 60 seconds. Raising it lowers engagement rate and breaks comparability with prior periods, so change it deliberately.