What is a good bounce rate in GA4, and why engagement rate matters more

There is no universal “good” bounce rate in GA4. Bounce rate is the share of sessions that were not engaged, so a good one depends on what the page is for, where the traffic comes from and how GA4 is set up. The useful questions are how you compare with similar sites (GA4 can show you that) and whether your own number is moving.

Target ranges you find online often come from vendor studies with their own samples, and some date from the Universal Analytics era, when bounce rate meant something else. Here is how to set a baseline you can defend instead, and why engagement rate is the number to watch.

What bounce rate means in GA4

Google defines it plainly: “The bounce rate is the percentage of sessions that were not engaged” (Google’s documentation). A session counts as engaged if it does at least one of three things:

1. Lasts longer than 10 seconds.
2. Has a key event (what GA4 used to call a conversion).
3. Has 2 or more page or screen views.

Engagement rate is the share of sessions that were engaged. The two always add up to 100%. If 1,000 sessions include 620 engaged ones, engagement rate is 62% and bounce rate is 38%. Google’s Traffic acquisition help page says it directly: “Engagement rate is the inverse of bounce rate” (Google’s documentation).

Why your old bounce rate benchmarks don’t apply

In Universal Analytics, a bounce was “a single-page session on your site”, and bounce rate was single-page sessions divided by all sessions (Google’s legacy UA article). Someone who read your whole pricing page for three minutes and left counted as a bounce, unless the page sent another hit such as an event.

In GA4, that same visit is engaged (it lasted more than 10 seconds), so it is not a bounce. The definition changed, so the numbers changed. A bounce rate target set in the Universal Analytics days, or a benchmark article that mixes both eras, will tell you nothing about your GA4 number.

Even Google’s old article refused to give a target. Its answer to “what is a good bounce rate” was “It depends”: a high rate is a problem when the site needs people to view several pages, and normal for sites where single-page sessions are expected, such as a blog (Google’s legacy UA article).

How to find bounce rate in GA4

Many standard reports don’t show bounce rate by default. An Editor or Administrator can add it (Google’s documentation):

1. Open Reports and go to the detail report you want, such as Pages and screens.
2. Click Customize report in the top right corner.
3. Under Report data, open Metrics and click Add metric.
4. Search for Bounce rate. If it does not appear in the list, it is already in the report.
5. Click Apply, then Save your changes.

You often don’t need to. Reports > Acquisition > Traffic acquisition already lists engaged sessions and engagement rate per channel (Google’s documentation), and since one is the inverse of the other, it gives you the same information. If you are new to these reports, our guide to seeing your website traffic in GA4 walks through them.

Use GA4’s own benchmarks instead of a blog post’s

GA4 has a benchmarking feature that compares your property with a peer group of similar businesses, and both bounce rate and engagement rate are among the metrics it covers (Google’s documentation). It uses GA4’s own definition and a peer group from your industry, which most published benchmarks cannot claim, so it is the one worth reading.

How it works, per Google’s help page:

• Where: the overview card on the Home page. Pick the metric in the card, expand the Benchmarking category, and turn benchmarks on with the badge in the card’s top right corner.
• What you see: your trend line, the peer group median as a dotted line, and a shaded range from the 25th to the 75th percentile.
• Who your peers are: your property’s industry category, taken from the industry you picked at setup plus signals such as your URLs. You can change the peer group from the same badge.
• Requirements: the Modeling contributions & business insights setting must be on in Admin, under Account settings. The peer group needs enough properties and your property needs enough users and data. Benchmarks don’t show when the date range is set to today, and the data refreshes every 24 hours.

Read it as a sanity check, not a target. Well below the shaded range, look at your traffic sources and landing pages. Inside or above it, the useful work is in your own trend.

Set your own baseline, page by page

A site-wide bounce rate averages very different visits. A blog post that answers a question on one page, a product page meant to lead to the cart and a contact page that should end in a form all behave differently. Mixed together, a change in the traffic mix can move the average while no single page got worse.

A baseline you can trust takes four steps:

1. Split by page type. In Pages and screens (with bounce rate added) or by landing page, group your pages into a few types: articles, product or service pages, the homepage, conversion pages.
2. Split by channel. In Traffic acquisition, read engagement rate per channel. Each channel has its own typical rate, and a lower one can be normal for that channel rather than a fault on the page.
3. Take 8 to 12 weeks. Note the typical engagement rate per page type and channel over that window, and how much it moves from week to week. That spread is your noise.
4. Watch for moves outside the spread. A good bounce rate is one that is stable or improving against your own history. A five-point jump on a page that has been flat for three months deserves a look. The same jump on a page that swings every week does not.

Why engagement rate matters more

Both numbers carry the same information, so this is about how they read and what they push you to do.

It reads the right way up. A higher engagement rate is better. When you report “engagement rate up from 54% to 58%”, nobody has to translate. Bounce rate carries baggage from the old definition and makes people think of single-page visits.

It sits next to your traffic. Engagement rate is in Traffic acquisition next to sessions and key events, per channel. That makes it a quality check on the traffic you are paying for or earning. When sessions jump and engagement rate drops, the new traffic is probably low intent: a campaign aimed at the wrong audience, a viral post, or bots. For bots, see the GA4 spike from China and Singapore, where engagement near zero is the tell.

It points to engaged sessions, the count that matters. A rate can rise because bad traffic disappeared while the good traffic stayed flat. Read engaged sessions next to it so you know whether you gained real visits or only lost poor ones.

This is why engagement rate made our list of marketing KPIs worth checking weekly and bounce rate did not.

Should you change the 10-second threshold?

The 10-second condition can be changed. In Admin > Data streams, pick your web stream, open Configure tag settings, click Show all, then Adjust session timeout. The field Adjust timer for engaged sessions sets how many seconds a session needs to count as engaged (Google’s documentation).

Raising it makes “engaged” stricter, which can make sense for long-form content where 10 seconds is barely a glance. The trade-offs:

• Your trend breaks. The rule that classifies sessions changes, so engagement rate and bounce rate move for that reason alone. Write down the date and only compare periods with the same setting.
• Peer benchmarks are likely to become less comparable, since your engaged sessions may no longer follow the same rule as most of your peer group.
• The other two conditions still apply. A session with a key event or a second page view stays engaged whatever the timer says.

For most small sites, leave it at 10 seconds and spend the effort on the pages and channels that pull the rate down.

What to do when engagement rate drops

Go through these in order before you touch a page:

1. Check the traffic mix. Did one channel grow fast? Its engagement rate pulls the average toward it.
2. Check countries and sources. A burst of sessions with near-zero engagement from one country or referrer is usually bots, not visitors.
3. Check the landing pages. Sort by sessions and look for one page whose engagement fell. A broken form, a slow page or a changed layout shows up there first.
4. Check tracking. A consent banner change or a tag problem can change how sessions and events are recorded. Our traffic drop diagnosis checklist covers the tracking checks step by step.

Getting the movement without checking GA4

The baseline approach works, but it relies on you remembering to open GA4 and compare weeks. Dialytics, our product, does that reading for you. It posts a weekly digest in Slack every Monday that compares the past week with the one before, engagement rate included, and comments on what moved rather than listing every number. A separate daily check sends an alert only when traffic or key events are unusual for that day of the week. You can mention the bot in the thread to ask a follow-up, such as which channel pulled engagement down.