What is a conversion rate? What the number actually means, and six ways it misleads

The short answer

A conversion rate is conversions divided by visits, times 100. If 45 of 1,800 sessions end in a completed enquiry form, your conversion rate is 2.5%. The formula is simple. The trouble is that "conversion" and "visit" are both definitions somebody chose. That is why WordStream reports an 8.18% average for Google Ads while Littledata reports 1.4% for the average Shopify store, and both can be right.

How do you calculate a conversion rate?

To calculate a conversion rate, divide the number of conversions by the number of sessions or users in the same period and multiply by 100. Use the same date range, the same traffic sources and the same definition of a conversion on both sides, or the result is two unrelated numbers divided.

A conversion rate is conversions divided by something. That is all it is. Every argument about conversion rates is really an argument about one of those two halves.

The top half is whatever you decided counts as a conversion. A form submission. A thank you page view. A click on a phone number. A file download. A live chat opened. Businesses routinely count different things and then compare the results, which is a bit like comparing your weight in kilograms to someone else's in pounds and concluding you are winning.

The bottom half is whatever population you divided by. All sessions. All users. Sessions from one channel. Sessions that reached a particular page. Sessions excluding bounces, which is a choice some tools make for you.

Change either half and the number moves, without one visitor behaving any differently.

What should count as a conversion?

A conversion should be the action that tells you a real person wants to talk to you or buy from you: a sent enquiry, a completed purchase, a call from the site. Write that definition down, then check the tracking matches it. Three things go wrong here more often than anything else.

• The event fires on the wrong action. A conversion recorded when the form comes into view, rather than when it is sent, counts curiosity as commitment. Rates built this way look healthy and produce no enquiries.
• One person counts twice. Someone who submits, gets an error, and submits again is two conversions and one enquiry. Over a small monthly volume, a handful of these visibly lifts the rate.
• Nobody has checked that the enquiry arrives. The event can fire perfectly while the email fails silently. Test this by sending one yourself from an address outside the company, and confirm it reaches the inbox a human actually reads.

A tracked event is not necessarily a valid enquiry. That distinction is the whole of it. In GA4 the setting that decides the top half is the key event, and our guide to GA4 key events for lead generation shows how to set one that fires on the send, not the view.

Which visits should you divide by?

Divide by the sessions or users that could reasonably have converted, from the sources you are judging, over the same period. Then write down what is filtered out. The denominator drifts more than people expect, and it drifts without anyone touching the website.

• A new consent banner means some visitors are never measured at all, so your denominator shrinks and your rate goes up.
• A bot filter turned on or off changes the denominator overnight.
• A campaign sending a lot of low intent traffic inflates the denominator, so the rate falls while the number of enquiries is flat or rising.

That last one matters. If you are judged on rate rather than volume, running a successful awareness campaign will make you look worse.

Here is the consent banner effect with the arithmetic shown. The figures are illustrative.

Nothing improved. A quarter more "performance" appeared because the bottom half shrank. The same trap sits under bounce rate, which is why a good bounce rate is harder to define than it looks.

Why do conversion rate benchmarks disagree so much?

Conversion rate benchmarks disagree because each one counts a different conversion, over a different population, on a different kind of site. A single-goal campaign page, a whole lead-generation website and an online store are not measuring the same thing, so their averages sit far apart.

None of these is wrong. They answer four different questions. For figures you can use with their limits stated, read what a good conversion rate is in 2026, by industry and channel, and for stores, the average ecommerce conversion rate.

Six ways the number misleads

The number misleads when its average hides the mix, its volume is too small, its definition quietly changed, it ignores lead quality, it is compared with someone else's benchmark, or it rises while revenue falls. Each has a simple check.

1. The average hides the mix

A site-wide rate of 2% can be one page at 8% and forty pages at nothing. The average is arithmetically correct and strategically useless. Always look at the rate for your top five landing pages separately before you look at the site total.

2. Small numbers move a lot

If you get fifteen enquiries a month, two extra is a 13% improvement and also completely normal variation. At small volumes, month on month percentage changes are mostly noise. Use longer periods, use absolute counts alongside the rate, and be sceptical of anyone celebrating a percentage without telling you the underlying count. The same maths is why most sites are too small for the tests described in A/B testing for marketing teams.

3. The definition changed under you

Someone renamed an event, added a second form, or set a new key event in the analytics property. The line on the chart steps up or down and it looks like a business change. Before investigating a sudden move, ask what changed in the tracking on that date.

4. It says nothing about quality

Twenty enquiries at a 4% rate, of which two are qualified, is worse than eight enquiries at 1.5% of which five are qualified. A form that is easy to complete because it asks almost nothing will raise your conversion rate and lower your sales team's opinion of marketing. Track the rate alongside qualified leads from the same cohort, or you will optimise your way into more work and less revenue.

5. The benchmark is not yours

You will be offered industry benchmarks. Treat them with suspicion, for a specific reason rather than general cynicism: most of them do not state what counted as a conversion, which countries the sample covers, what business model it covers, or how many businesses are in it.

When we looked for a benchmark we could publish responsibly for Australian B2B professional services, none of the eleven sources we examined defined the metric the way we needed and permitted redistribution. That is why our methodology works from your own numbers, not a borrowed benchmark. A rate is only comparable to another rate built the same way, and you almost never know how the other one was built.

6. A better rate can still mean less money

The rate is the middle of a chain: visitors, enquiries, qualified leads, won work, contract value. A change that lifts the second link and damages the fourth leaves you worse off with a nicer chart. Whenever you report a rate improvement, report what happened to won work in the same period, even when it is inconvenient.

So what is a good conversion rate?

A good conversion rate is one that is rising on your own site, with the definition held still, while qualified leads rise with it. Published averages give you a rough range, not a target, because they rarely count what you count.

Nobody can tell you honestly what yours should be, and you should be wary of anyone who does without asking what you count. The useful comparison is your own site against itself, over time. Second best is one page against another page on your site, same definition, same period. Everything else is conversation, not evidence.

Once the number is trustworthy, the work of raising it is conversion rate optimisation, and it starts with a CRO audit rather than a redesign.

How do you make your conversion rate trustworthy?

You make it trustworthy by writing down the definition, testing that it fires, recording the denominator and dating the note. It takes less than an hour and none of it is technical.

1. Write the definition down. "A conversion is a completed submission of the enquiry form on any service page, counted once per session." One sentence, in a document other people can find.
2. Test it. Submit the form. Confirm the event fires, the count increments, and the email arrives.
3. Record the denominator. Which sessions, which sources, what is filtered out, and what the consent banner does to it.
4. Note the date you did this. When the number moves in four months, the first question will be whether anything changed, and you will have the answer.

That document takes twenty minutes and makes every future conversation about performance shorter. A full GA4 audit checklist covers the rest of the property if you find more than one problem.

PostCliq, a free website audit, checks what sits in front of the top half: whether the forms and calls to action on your public pages are there, visible and working the way a visitor would meet them. A free audit of your website shows where visitors are likely to stop before they ever become a conversion.

What to do next

1. Find your current conversion rate and the date range it covers.
2. Write, underneath it, the sentence that defines the top half and the bottom half.
3. Submit your own form and confirm the event, the count and the email all happen.
4. Look at your top five landing pages separately before you look at the site total.
5. Report the rate next to qualified leads from the same period, every time.

If you cannot write the definition in step 2 without checking with somebody, you have found this quarter's most valuable half hour. When you are ready to turn it into a conversation with your director, the director conversation worksheet takes you from the rate to the five lines that matter to them. For the bigger picture of why the visit after the click now matters so much, read what post-click optimisation is.

Frequently asked questions

What is conversion rate in simple terms?

Conversion rate is the percentage of visitors who do the thing you want them to do. If 100 people visit your site and three send an enquiry, your conversion rate is 3%. The simple version hides two decisions: what you count as "the thing", and which visitors you count. Write both down before you compare your rate with anyone else's.

How do you calculate conversion rate?

Divide conversions by sessions or users for the same period and multiply by 100. For example, 45 enquiries from 1,800 sessions is 45 divided by 1,800, which is 0.025, or 2.5%. Use matching date ranges and sources on both sides, and count each person's enquiry once, or the percentage will not mean what you think it means.

What is the difference between conversion rate and sales conversion rate?

A website conversion rate usually measures visits that become enquiries or orders. A sales conversion rate measures leads or opportunities that become customers, so it sits later in the chain. Both matter. A marketing team that lifts the first while the second falls has usually made the form easier and the leads worse, so report them side by side.

What is a good conversion rate for a website?

There is no single good number. Ruler Analytics reports a 5.13% average across 13 lead-generation industries in 2026, Unbounce a 6.6% median for landing pages, and Littledata 1.4% for the average Shopify store. Each counts something different. Use them as a rough range, then judge your site against its own history with the definition held still.

Why did my conversion rate suddenly change?

A sudden step up or down is usually a tracking change rather than a business change. Check first for a renamed or new key event, a second form, a consent banner, a bot filter or a new campaign sending different traffic. If none of those changed on that date, look at the landing pages and forms next, starting with whether the form still sends.