What Is a Good Bounce Rate?

GA4 flags your bounce rate at 61% and offers nothing else. No verdict, no comparison, no note about whether that's fine. Just a number sitting there like a dashboard light with the manual torn out of the glovebox.

There's no single figure that counts as a good bounce rate everywhere. A blog post bouncing at 65% and a checkout page bouncing at 20% can both be healthy, because the metric describes what happened on one page, not whether the business behind it is working. It's one line inside a bigger stack of traffic numbers worth checking before you credit or blame any channel for it.

Then GA4 changed what the number even measures, and plenty of sites watched their bounce rate move overnight with nothing else different about the traffic.

The question worth answering isn't whose benchmark you match. It's which page, which traffic source, and which definition of "bounce" you're actually looking at. Start there.

What Counts as a Bounce in GA4 (It's Not What You Think)

Universal Analytics called a session a bounce if the visitor viewed exactly one page, no matter how long they stayed on it. Read for eleven minutes and left from the same page you landed on? Bounce.

GA4 threw that definition out. A session now counts as engaged if it crosses any one of three thresholds, not all three.

Google Analytics Help lays out the exact wording: bounce rate is simply the inverse of engagement rate. A session only counts as a bounce when it misses all three at once: under ten seconds, no key event, and only the one page viewed.

That changes the arithmetic. Say a property logs 1,000 sessions in a week and 454 of them never cross any of the three thresholds. The bounce rate for that week is 45.4%, and the other 546 sessions make up the 54.6% engagement rate sitting right next to it.

There's a failure mode baked into the new definition. If a low-value interaction, a scroll past 25% of the page, say, is marked as a key event in GA4, every session that trips it instantly reads as engaged, even if the visitor left four seconds later having scrolled once out of boredom. Check Admin > Events > Key events before trusting the number. A key event tied to something that fires on nearly every visit quietly erases your bounce rate without anyone touching the actual page.

It also means a historical comparison against old Universal Analytics data is close to meaningless. The same traffic, measured the old way and the new way, produces two different bounce rates, and neither one is wrong. They're measuring different things.

What a Good Bounce Rate Looks Like by Page Type

Here's where most advice stops short: it hands you one number for the whole site. The honest version splits by what the page is actually for.

Those ranges come from Plausible's 2026 industry benchmarks. A B2B SaaS site specifically runs narrower: marketing content lands around 50% to 65%, deep tutorial pages drop to 35% to 50%, and product or pricing pages sit at 35% to 55%, according to daydream's 2026 benchmark breakdown.

Run the table against a real page before you trust it. Say your SaaS pricing page sits at 58%. Against the all-site median, that looks troubling on its own. Against the product-page range of 35% to 55%, it's three points over the top, borderline but not alarming. Against last month's number for that same page, which is the comparison that actually matters, it might be flat, or it might be climbing. The industry range tells you where you stand in the field. Your own trend line tells you whether anything changed.

Zoom out past page type and the cross-industry median still holds up as a sanity check. Databox's September 2024 survey put the median bounce rate across all industries at 44.04%, with SaaS specifically averaging 48.27%, as reported in Semrush's bounce rate guide.

Contentsquare's 2025 Digital Experience Benchmarks report adds a simpler rule of thumb on top of all that: 40% or under is generally considered good, 55% or higher is high enough to look into. Treat it as a starting filter, not a verdict, until you've checked it against the page type above and against the page's own history.

Benchmarks by Where the Traffic Came From

Page type explains part of the spread. Where the visitor came from explains another part, and it's the one most people skip.

digitalapplied's 2026 channel data breaks bounce rate down by search intent instead of industry, and the gap is wide.

That's a 17-point spread driven entirely by how pre-qualified the visitor already was before they clicked. Judging a page's bounce rate without knowing which of those three categories sent the traffic is the same mistake as judging a channel by impressions instead of clicks: the raw number hides which audience it actually describes.

The same report gives percentile breakpoints worth keeping on hand: the median site sits at 47.4%, better sites sit at 36.1% or under, and the strongest 10% sit at 27.8% or under.

A site running both branded search and a cold outbound campaign pointed at the same landing page often shows one blended bounce rate in the aggregate report, hiding the fact that branded clicks might bounce at 30% while the cold clicks bounce at 60%. Split them and the read flips from "the page is fine" to "half this campaign's spend isn't landing."

Segmenting by source means tagging campaigns so GA4 can actually split them out. If every link just says "newsletter" or "Reddit" in your head with nothing in the URL to prove it, tag it with a UTM code before running this comparison again next month. The same logic applies to organic traffic: a page ranking for ten different queries can have ten different real bounce rates hiding inside one average.

How to See Your Actual Bounce Rate in GA4

GA4 doesn't put bounce rate on a report by default. You have to go looking for it.

1. Open Reports > Engagement > Pages and screens. This gives you per-page data instead of a site-wide blend.
2. Click the pencil icon in the top right to customize the report.
3. Under Metrics, add Bounce rate. It now sits as a column next to engagement rate and average engagement time for every page.
4. Add Session source/medium as a secondary dimension to see the same page split by channel instead of blended across all of them.
5. Export or screenshot the result. GA4's customizations don't always persist across browser sessions the way you'd expect, and rebuilding the same view twice a week wastes time you don't have.

Once the column is there, read bounce rate next to average engagement time, never alone. A page pulling 2,000 sessions a month at a 70% bounce rate with four seconds of average engagement time has a real problem: people are leaving almost immediately, probably before anything on the page even finished loading. The same 70% bounce rate paired with 90 seconds of average engagement time tells a different story. That visitor read the page, got the answer, and left satisfied.

Same percentage. Opposite diagnosis. The engagement-time column is what tells you which one you're actually looking at, and it's worth pulling for every page before you touch anything on the site.

One more timing note worth building into the habit: GA4's standard reports run on a short processing lag, so a page that looks alarming at 9am on a Monday might just be showing an incomplete weekend. Check a full, closed week before deciding a number means anything.

When a High Bounce Rate Is Actually Fine

Some pages are built to end the visit, not extend it. A contact page with a phone number on it is bounce-worthy by design: the visitor got the number, closed the tab, and dialed. GA4 logs that as a bounce unless they lingered past ten seconds first, which a visitor who already has what they came for rarely does.

A glossary entry or a single-question page like this one works the same way. The visitor arrived wanting one answer. If the page gives it to them cleanly in under ten seconds and they leave, the page did its job. A 70% bounce rate on that kind of page next to a six-second average engagement time usually means people are reading the whole thing fast, not skipping it.

"Some pages exist to satisfy intent quickly. When they succeed, users leave fast. Not because they're disengaged, but because they got exactly what they came for."

That's Dmytro Sokhach, co-founder of Editorial.Link, quoted in Semrush's bounce rate guide. It's the single best test for whether a high number is a problem: did the visitor leave because the page failed, or because it finished the job in record time?

Chasing a lower bounce rate on a page like that usually means adding friction that wasn't needed. A popup demanding an email before the answer loads is the most common offender, and a second page forced into what used to be a one-page journey is close behind it. Both will drag the number down. Neither will make the page better, and the second one will show up again in a worse engagement-rate-per-conversion number a month later.

What to Do When It Actually Is a Problem

A genuinely bad bounce rate has a short list of usual suspects, and most of them are fixable inside a week.

1. Check load time first. If the largest content on the page takes longer than 2.5 seconds to render, a share of visitors are gone before they ever see it. Slow is indistinguishable from broken to someone who just clicked an ad.
2. Match the promise to the page. If a paid link or ad headline says "free trial" and the landing page opens on a pricing table with no trial button above the fold, that mismatch is the bounce. Fix the headline or fix the page; don't leave the gap between them standing.
3. Add exactly one internal link above the fold, pointed at the next logical question a satisfied reader would have. One obvious link beats five competing ones almost every time.
4. Check the mobile view specifically. Mobile bounce rates run consistently higher across nearly every benchmark cited above, and a layout that reads fine on a 27-inch monitor can be unusable on a 6-inch screen worked with one thumb.
5. Wait fourteen days before calling it fixed. Bounce rate swings with day-of-week traffic mix on its own; a four-day sample after a change will lie to you in both directions, sometimes in the direction you were hoping for, which makes it worse.

This is the same floor that conversion rate optimization work starts from: fix what's broken before testing what's merely average. A page nobody stays on long enough to read never gets the chance to convert anybody, no matter how good the offer underneath it is.

What Bounce Rate Tells You About a Traffic Source You're Paying For

Every paid channel eventually has to answer the same question: did that click actually do anything once it landed.

A click that bounces in four seconds still shows up in the campaign report as a click. It still cost whatever it cost. Running that number against what CAC actually measures before declaring a channel worthwhile catches a lot of spend that looked fine inside the ad platform and looked very different once the landing page's engagement time showed up next to it.

That check applies to any paid placement, not just search ads.

The Board is a leaderboard where rank is bought outright, in dollars, nothing else, disclosed as exactly that every time it's mentioned here. Every listing's impressions and clicks are public for the same reason this whole post exists: so the brand running it can pull those clicks into GA4 by source and medium and check whether the people who clicked through stuck around or bailed in three seconds. A listing that drives traffic nobody reads past the first ten seconds is telling its owner something different than the rank number alone ever will.

Pull up GA4 tomorrow, split your top landing pages by session source, and put average engagement time in the column right next to bounce rate before you change anything. The percentage on its own was never the metric that mattered. What it's sitting beside always was.