How to Lower Your Bounce Rate
To lower your bounce rate, first work out whether it is real. Most alarming bounce rates turn out to be a tag firing twice, a consent change, or a campaign sending visitors the page was never built for. Once the measurement is clean, segment by landing page, source and device, then close the intent gap between what the visitor was promised and what the first screen delivers. That gap, not the page design, is what sends people back to the results. And accept that some pages should have a high rate: a delivery information page that answers the question on the first screen has done its job.
- Most sudden bounce rate changes are measurement changes, not behaviour changes.
- A blended bounce rate has no single cause, so it cannot have a single fix.
- The intent gap between the promise and the first screen explains more bounces than design does.
- Some pages are supposed to bounce: a completed task and an abandonment look identical.
- Measure the next step you actually wanted, per template, rather than a site wide rate.
Bounce rate is the share of visits that ended on the page they started on, and it is the most confidently misread number in eCommerce analytics. It went out of fashion, came back under a new definition, and in the meantime accumulated a decade of advice written against the old one. Lowering it is easy and frequently pointless. Working out whether yours is telling you anything is harder and worth the afternoon. Last updated: September 2026.
Omniconvert has measured how storefronts present their first screen across the CROBenchmark dataset of 7,000+ websites in 15+ industries, against 248+ audit criteria, over 13 years in eCommerce. The pattern in that data is not that high bounce pages are badly designed. It is that they are usually answering a different question from the one the visitor arrived with, and that the teams who improve fastest are the ones who checked their tracking before they redesigned anything.
What follows is the method: what the number counts now, why most high rates are not a problem, the Bounce Triage for finding the ones that are, and what to put on the dashboard instead. If the underlying question is really about revenue rather than this one metric, a good conversion rate is the better place to start, and our guide to how to improve your conversion rate covers the levers in priority order.
What a bounce rate actually measures now
The practical consequence is that your bounce rate is a statement about your measurement configuration at least as much as it is a statement about your visitors. Two stores with identical traffic and identical pages can report rates twenty points apart because one counts a ten second read as engagement and the other does not.
This is not a reason to ignore the metric. It is a reason to pin the definition before you compare anything. Write down what counts as engagement on your property, what the dwell threshold is, and which events are firing. Until that is on paper, every comparison you draw, against last month or against an industry figure, is a comparison of two different measuring instruments.
The second thing to pin is the denominator. Sessions are not a stable unit. A consent banner change, a redirect chain, an app update, a bot wave: each of these changes how many sessions exist without changing how many people visited. A rate whose bottom half moves on its own will move on its own, and it will look like a behavioural finding.
Why most high bounce rates are not a problem
Three categories of page will always report a high rate and should.
Task pages. Store locators, delivery and returns information, order tracking, size guides, contact details. The visitor arrives with one narrow question, gets the answer on the first screen, and leaves. That is the page working perfectly. Redesigning it to force a second pageview makes the experience worse and the number better.
Well answered content. An article that answers its title question in the opening paragraph will lose readers who only wanted that answer. The alternative, burying the answer to drive scroll depth, is the pattern that answer engines and readers both punish. Statista's surveys of online information behaviour have consistently shown how quickly people move on once they have what they came for.
Single intent campaign traffic. A promotion sent to an email list arrives with a specific purchase in mind. A large share will buy or leave, with no browsing in between. High bounce, healthy revenue.
Set those aside and the remaining pages are where the metric earns its keep: category and product templates reached from paid traffic, and landing pages built for a campaign. Those are the ones to triage.
The Bounce Triage: four checks in order
One: rule out the measurement. Check for a second analytics tag firing an extra pageview, which mechanically halves your reported bounce rate and has fooled a great many teams into celebrating. Check whether consent mode is suppressing the events that would mark a session engaged. Check for redirects adding a hop. This step costs an hour and resolves a surprising share of sudden changes.
Two: segment before you diagnose. Split the rate by landing page, then by traffic source, then by device. A blended figure is an average across pages that have nothing in common, and the average will hide one template performing badly behind several performing fine. Almost every real bounce problem is a single segment, and the segment view names it without any further analysis.
Three: close the intent gap. Open the ad, listing or email that sent the traffic, then open the landing page beside it, and read the first screen only. Does the page confirm what the visitor was promised, in the same words, without scrolling? A campaign promising a specific discount that lands on a generic category page has an intent gap, and no amount of page polish will close it. Either the page changes or the targeting does.
Four: give the page one obvious next step. After the promise is confirmed, the visitor needs somewhere to go. One continuation, above the fold, specific rather than decorative. Baymard Institute's long running checkout and navigation research has documented how reliably clarity at the point of decision shapes what happens next, and the first screen of a landing page is a point of decision.
Speed belongs inside this sequence rather than at the top of it. A site wide speed problem would show as a site wide pattern, which is rare. A template specific one, on mobile data, with a heavy hero asset or a blocking third party script, is common and shows up clearly once you have segmented by device. Shopify's own platform reporting has made mobile the majority of storefront sessions for years, so a mobile only bounce problem is a majority problem.
What the number cannot fix, and what to measure instead
The table below is the substitution we recommend in audits. It is deliberately per template, because that is the level at which a bounce is either a problem or a completed task.
| Page type | What a high rate usually means | Measure this instead | Worth fixing? |
|---|---|---|---|
| Paid landing page | Intent gap against the ad | Next step rate from the first screen | Yes, first |
| Product page | Missing decision facts | Add to cart per session | Yes |
| Category page | Findability or slow first render | Product view per session | Yes |
| Blog or answer page | Question answered on screen one | Scroll past the answer, then onward click | Rarely |
| Delivery, returns, tracking | Task completed | Task completion, or support contacts avoided | No |
| Home page | Traffic mix change | Entry to category rate | Check source first |
Read the right hand column as the actual brief. A paid landing page with a poor next step rate is a real problem with a known owner and a testable fix. A delivery page with a high bounce rate is not a problem at all, and the hours spent on it are hours not spent on the first row.
What a store owner should do this week
- Check for a duplicate analytics tag. Twenty minutes in your tag manager and your page source. If you find one, your historical bounce rate is not comparable and you should say so before anyone draws a trend from it.
- Segment by landing page, source and device. One afternoon. Write down the three worst templates. In most stores the list is short and one item dominates it.
- Read your worst paid landing page against the ad that feeds it. First screen only, on a phone. Ask whether the promise is confirmed in the same words. This single comparison finds more recoverable revenue than any redesign.
- Replace the metric on one dashboard. Swap the site wide bounce rate for next step rate on your three main templates. A store that wants that prioritisation done across every template rather than three can see how Nexus by Omniconvert unifies commerce data and ranks experiments by True Profit, with a human approving what goes live.
If the triage points at the checkout rather than the landing page, that is a different fix list and a longer one. To work through the whole storefront in order instead of one metric at a time, our companion guide will audit it step by step, and the controlled testing half of the work sits with Omniconvert Explore, the CRO platform that averages a 23.2% conversion uplift across 70,000+ experiments precisely because those are controlled tests rather than before and after snapshots.
Three ways a lower bounce rate can make things worse
First, pagination. Splitting one useful page into five reduces bounce to near zero and makes the content worse for everybody, including the answer engines that now supply a growing share of pre purchase research.
Second, interstitials and modals. A popup that fires before the visitor has read anything will register as engagement and will cost you the visit. The metric improves, the session ends anyway, and now it ends without the visitor having seen the offer.
Third, automatic events. Marking a session engaged on a timer or a scroll event that fires on load technically works. It also permanently destroys the only diagnostic value the metric had, and it is not reversible for historical data.
The common thread is that bounce rate is easy to move and hard to move honestly. Nielsen's long standing work on how people read online is a useful corrective here: visitors scan, decide quickly, and leave when the page does not match what they came for. The fix for that is a page that matches, not a page that detains.
The bottom line
Stop trying to lower your bounce rate and start asking which pages it is telling you about. Run the triage in order: confirm the measurement, segment the rate, read the landing page against the promise that sent the visitor there, and give the first screen one obvious next step. That sequence costs a day, needs no budget, and will usually find that two templates out of twenty account for most of the recoverable loss while the rest of the high bounce pages were doing exactly what they were built to do. Then replace the site wide figure with the next step you actually wanted on each template, and the metric stops being a number people argue about and becomes a brief somebody can act on.