Traffic but no sales: a 20-minute diagnostic

Two thousand visitors. Four sales. Something is wrong and nobody will tell you what.

When a site has traffic but no sales, the cause is rarely mysterious. It is usually one group of visitors dragging your average down, and you can find which one in about twenty minutes by reading three splits in your analytics before you change a single word on the page.

Most advice on this starts with your headline. That is the wrong end. Start with the denominator.

Why does a site get traffic but no sales?

A site gets traffic but no sales when one group of visitors drags the average down far enough to hide the fact that another group converts fine. The three splits that expose it are human versus automated traffic, mobile versus desktop, and new versus returning visitors. Check those three before anything else.

Your conversion rate is a fraction. Everyone stares at the top of it, the sales. The bottom of it, the sessions, is where the lie usually lives.

How do you check yours?

Pull the same conversion rate three times, split by device, by new versus returning, and by traffic source. Then look at sessions by city. If one of those numbers is wildly out of line with the others, you have your suspect. The whole thing takes about twenty minutes.

1. Sessions by city. If Ashburn, Council Bluffs or Singapore sit near the top and you sell to nobody there, some of that traffic is data centre traffic, not customers. Your real rate is higher than you think.
2. Conversion rate by device. A gap wider than the 74% Contentsquare records points at mobile, not at your product.
3. Conversion rate by source. If organic converts and paid does not, the page is fine and the targeting is not.
4. New versus returning. If both are near zero, the problem is on the page. If returning buys and new does not, you are losing people in the first ten seconds.

Three of those four checks can clear your website of blame entirely.

Paste your URL and see these splits scored automatically.

What is going wrong underneath?

People decide whether to stay in roughly ten seconds, before they read anything below the fold. Chao Liu, Ryen White and Susan Dumais at Microsoft Research analysed dwell time across 205,873 web pages and more than two billion visits, and found the risk of leaving is highest in the opening seconds and flattens after about thirty.

Nielsen Norman Group's write-up notes that 99% of pages show negative aging. People leave fast or stay long, with little in between. Your homepage is not being read. It is being triaged.

If you cannot say what you sell, who it is for and why it beats the obvious alternative inside ten seconds of screen time, the rest of the page never gets a vote.

How strong is the evidence here?

Strong on the pattern, weak on your specific number. Large datasets agree that traffic quality and device explain most of the variance between sites. No benchmark can tell you what your own site should convert at, because published averages mix a 5% food brand with a 0.7% jewellery brand and report something in the middle.

Conversion rate by visitor type and channel, from the Contentsquare benchmark:

Read the bottom row again. A site running mostly organic social traffic at 0.8% is normal. The same rate on paid search is a fire.

For what happens after someone decides to buy, the Baymard Institute's running average of 50 studies puts cart abandonment at 70.22%. Among those abandoning for a fixable reason, 40% blame unexpected extra costs and 18% refuse to create an account. Baymard sells UX audits, so weigh their framing accordingly. Imperva is a security company with no stake in the CRO answer.

Get that arithmetic run across your whole site, free.

The fix, step by step

Fix the measurement before the page. Filtering fake traffic changes every number the later steps depend on, so doing it second wastes the work. After that, the order runs cheapest to most expensive: the first screen, then the phone, then the checkout, then the ad match.

1. Filter the fakes first. Exclude known data centre traffic and your own visits, then recalculate. Everything downstream depends on this number being real.
2. Fix the ten second answer. At 360px, above the fold: what it is, who it is for, why it beats the alternative, and one thing to click. Nothing else.
3. Close the device gap. Buy something from your own store, on your own phone, on mobile data. Most mobile problems are found this way rather than in a report.
4. Kill the checkout surprises. Show shipping and tax before the last step, and let people buy without an account. Those are two of the largest fixable abandonment reasons Baymard records.
5. Match the page to the ad. If the ad promises a price and the page opens on a brand story, you paid for a bounce.

Did it work? How to tell

Track conversion rate by device and by source rather than the blended site average, and compare four weeks against the four weeks before you changed anything. You need roughly 300 conversions across that comparison to trust a difference smaller than 20%. Below that volume, judge the specific drop off you targeted instead.

If you cannot reach that number, do not pretend a 12% lift is real. Watch the step you changed, and read how many form fields is too many before cutting at random.

Who should do the opposite

Two cases make this diagnostic a distraction. Under about 1,000 sessions a month, every split above is noise and your problem is demand rather than conversion. And if all your traffic comes from one paid campaign launched last week, you do not have a site problem yet, you have an audience you have not found.

There is a ceiling on our own evidence too. Revslip has audited 134 businesses across roughly 200 conversion signals, and those sites belong to owners who already suspected something was broken. That makes the set worse than average by construction. It shows what tends to co-occur. It is not a population sample.

Questions people ask when nobody is buying

Three questions come up more than any others once the sales stop: whether the conversion rate is genuinely bad, whether a redesign is the answer, and what it costs to pay someone to find the problem. Short answers below, with the longer arguments linked out.

Is a 1% conversion rate bad?

It depends on your traffic mix. At 1% on paid search, something is badly wrong. At 1% on organic social you are above the 0.7% Contentsquare records for that channel. The blended number tells you almost nothing alone.

Should I redesign the site?

Rarely as a first move, because a redesign changes everything at once and measures none of it. Work through whether you need a website redesign before committing budget.

What does it cost to have someone find this?

Agency and consultant bands vary more than people expect, and below a certain traffic level none of them pay back. The numbers are in what a CRO audit actually costs.