A high bounce rate isn't always bad: two kinds of bounce

Almost every team that starts looking at analytics gets the same question at some Monday meeting: "Why is our bounce rate so high?"

Someone goes looking, finds advice that bounce rate "should be under 40%", and starts reworking the home page, adding pop-ups and stuffing articles with recommended reading. A month later the bounce rate is a few points lower and conversion hasn't moved.

The problem was there from the start. Bounce rate is a blended metric. It puts two completely different kinds of visit into the same numerator.

How a bounce is counted

First, the definition. A session with exactly one pageview and no custom event is a bounce. Bounce rate = bounced sessions ÷ all sessions.

Notice that time plays no part. Someone who leaves after three seconds and someone who reads the whole article for six minutes and then closes the tab are both bounces.

Bounce one: they got what they came for

Picture someone searching "what do UTM parameters mean", clicking through to an article on your blog, reading it for five minutes, noting the answer and closing the tab.

From their point of view that was a very successful visit. From the analytics point of view it was a bounce.

Content sites, help centres, documentation and single-page tools naturally produce lots of these. We've seen well-run technical blogs sitting at 75–85% bounce, which is completely normal. Push that number down by filling articles with "you might also like" boxes and readers usually get more annoyed, not more engaged.

Bounce two: they didn't find it, or ran out of patience

Someone else taps your social ad and lands on a product page. It takes four seconds to load, the first screen is a big image and a slogan, they can't tell what you do, and they hit back.

That's a bounce too — but it's a click you paid for that bought nothing.

This is the kind worth fixing. It tends to cluster in:

• Landing pages for paid traffic, where the ad promises something the page doesn't show.
• Slow pages, especially on mobile, where people leave before the first screen appears.
• Mismatched sources, such as a referral that sends people to an unrelated page.

Telling them apart

Use engagement rate instead

Engagement rate is closer to "was this visit worth anything?". A session counts as engaged if it lasts at least 10 seconds, or has at least two pages, or hits an action you've marked as key.

Back to our two visitors: the reader who stayed five minutes is engaged; the one who left after four seconds isn't. Engagement rate separates them; bounce rate can't.

So our advice: watch engagement rate day to day and treat bounce rate as secondary.

Look at page types separately

Split content pages (blog, help) from conversion pages (landing, pricing, sign-up). Filter the report by path, or create two page groups — /blog/* and /lp/*, say.

A high bounce rate on content pages is no emergency. A high bounce rate on conversion pages deserves a closer look.

Split by source as well

On the same landing page, visitors from search and visitors from ads often behave very differently. Break the landing page down by channel:

(Sample data)

The table is specific: the page isn't the problem; the people the social ads bring don't match it. What needs changing may be the ad creative or targeting, not the page.

Make the key action a goal

If the landing page has one action you care about most — clicking "Book a demo", finishing the product video, scrolling to pricing — make it a goal and mark it as a key event. Single-page visits that hit it count as engaged, and your number moves closer to the real share of useful visits.

When bounce rate really is telling you something

Signals worth taking seriously:

1. A jump of ten or more points in a single day. Usually technical: a page error, the snippet installed twice or lost, or a channel suddenly sending lots of bots. Check the errors and bots reports first.
2. Mobile far higher than desktop. Open the performance report, group by device and look at largest contentful paint. Mobile loading is the usual suspect.
3. A rise after a redesign. Compare the same channel before and after the launch date. If it changed noticeably, the redesign may have hidden something people needed.

In short

Next time someone asks why the bounce rate is so high, ask three questions back: which kind of page, which channel did those visitors come from, and how long did they stay?

Answer those and you'll usually find that what really needs work is two or three pages and one or two channels — not one vague number.

Related articles

Analytics basics·Oct 1, 2026·3 min read

How ad clicks without UTM tags still get recognised

Ad platforms add a string of parameters to landing URLs on their own. Understand them and you'll see why "direct" traffic sometimes hides paid clicks — and how to get them back.

Read article

Product notes·Oct 16, 2026·3 min read

Why bot traffic gets its own report

By default we take identified bots out of your human reports rather than dropping them or mixing them in. The three reasons behind that, and what it means for your numbers.

Read article

Playbooks·Oct 14, 2026·5 min read

Checkout drop-off: a seven-step investigation from data to fix

Plenty of add-to-carts, not many payments. A checklist to work through in order — rule out technical problems first, then find which step, which visitors and what cause.

Read article