Is it your traffic or your website? The four-check split

If the traffic is fine and the site is broken, every euro you add to ads makes the problem bigger. Four cuts of data you already have will tell you which one you are dealing with.

You have had this argument. Maybe with a co-founder, maybe with yourself at eleven at night. The ads are running, people are arriving, nobody is buying. One side says the targeting is wrong. The other says the page is wrong. Both are guessing, and the guess costs money in opposite directions.

Is it my traffic or my website?

Split conversion by source before you touch anything. If every source converts badly, it is the site. If one source drags the average down while the rest hold up, it is that traffic. A single site-wide number cannot separate the two, which is exactly why the argument never ends.

That is the whole method. Everything below is how to run it without fooling yourself. If nothing at all is selling, run the traffic but no sales diagnostic first and come back.

How do you tell the difference?

Four cuts, ninety days of data, about twenty minutes. Conversion by source, engagement by source, branded against non-branded search, and new visitors against returning ones. Each isolates a different failure. Run all four and usually one answer is left standing.

1. Conversion by source. Same page, split by where people came from. Spread means the page works and one channel is sending the wrong people. Flat and low across everything means the page.
2. Engagement by source. Bounce and scroll depth per channel. People who leave before scrolling never saw your offer, so nothing on the page can be blamed for it yet.
3. Branded against non-branded search. If people searching your name convert and people searching your category do not, the page cannot sell to a stranger. That is a site problem wearing a traffic costume.
4. New against returning. Returning visitors converting fine while new ones do not points at the first screen, not the product.

The decision rule: all sources bad means the site. One source bad means that source. Branded fine and non-branded dead means your value proposition. Returning fine and new dead means your homepage.

Get the site half of the answer in minutes, free.

Why does the same page convert differently by source?

Because the visitor is different, not the page. Someone typing your name into a search box has already decided something. Someone stopped by an ad mid-scroll has decided nothing at all. The page has to do far more work for the second person, and most pages are written for the first.

Contentsquare's 2026 benchmark, drawn from 99 billion sessions across more than 6,000 sites, is unusually direct about this. It warns against treating paid social as an immediate conversion channel at all, on the grounds that those ads reach people early in their thinking. Worth noting the source sells analytics, and that this particular advice cuts against its own commercial interest.

Those two groups behave nothing alike, and averaging them produces a number that describes neither.

A site-wide conversion rate is the average of several different arguments. You cannot win an average.

How good is the evidence on traffic quality?

Better than it used to be, and more skewed than the headlines suggest. The Association of National Advertisers, a trade body for advertisers rather than a vendor, tracks how much money lands on junk inventory. The median has collapsed. The tail has not.

Share of ad spend landing on junk inventory. The median bar is invisible because the median is now almost nothing, and that is the actual answer to the question in the title. For most advertisers traffic quality has been cleaned up, so the site is the likelier culprit. For a minority it is severe. Assuming you are in the tail is expensive, and so is assuming you are not.

What to do once you know which one it is

Price the gap before you act on it. Take your best-performing source as the ceiling that page can reach, then work out what the underperforming ones would be worth at that rate. The formula is traffic × CVR gap × AOV × mobile weight, published openly so you can check it by hand.

Say paid social sends 9,000 sessions a month at 0.4 percent while organic converts at 2.2 percent on the same landing page, with a 60 euro order value. Closing even half that gap is roughly 48 orders, or 2,880 euros a month.

If the number is large, the page is the constraint and the traffic is only exposing it. If it is small, you are optimising a channel that was never going to pay, and the honest move is to stop buying it.

Paste your URL and see which side the problem is on.

Once you know it is the site, the ranked list of what usually breaks is in the conversion leak index. If someone has quoted you for the work, the price bands are in what a CRO audit actually costs.

How to tell if you fixed the right thing

Watch the underperforming source on its own, not the site-wide rate. A fix that works usually narrows the gap between your worst channel and your best one. If the average moves but the gap does not, something else changed and you are about to credit the wrong thing.

Give it four weeks and matched periods. Same weeks, same spend, same promotions. Write down what you expect before you look, because the number is easy to read backwards afterwards.

Who should ignore all of this

Anyone under roughly 500 conversions a quarter. Below that, channel-level rates bounce around so much that the spread you are staring at is noise, and you will confidently diagnose a problem that does not exist. Use judgement and the page itself instead.

Two limits on our side of this, stated plainly. Revslip measures the site half well and the traffic half not at all, because we crawl pages rather than sit inside your ad accounts. And the 134 sites we have audited are not a fair sample of anything: their owners came to us already suspecting a problem, which tilts our frequency numbers toward sites that had one.

Questions people ask about traffic quality

Three keep coming back. Whether bad traffic can be fixed with better targeting, what to do when the channels disagree, and whether any of it applies before you spend on ads at all.

Can I fix bad traffic with better targeting?

Sometimes, and it is the cheaper thing to try first. Tighter targeting on a channel that converts at a tenth of your others often recovers a good share of the gap. What it cannot fix is a page that only makes sense to people who already know you.

What if the channels disagree with each other?

Then you have your answer already. Disagreement is the signal. A page that converts well for one source and badly for another is a page that works, meeting an audience it was not written for.

Does this matter if I am not running ads yet?

More than it will later. Buying traffic to a page that cannot sell to strangers is the most reliable way to lose money quickly, and the branded against non-branded check works on organic search alone. If the answer points at rebuilding, read redesign or fix before committing.

How do I know if my conversion rate is bad at all?

Compare it to your own segments first, then to your sector rather than to a blended industry average. Sector rates ran nine times apart in a single month of measured data, which is covered in real 2026 benchmarks by sector.

Every source bounces. What is the page doing wrong?

Something in the first screen is failing a check that runs before anyone scrolls. Across 205,873 pages the risk of leaving is highest on arrival and falls after, so the opening carries it. Why people leave a landing page covers the four checks.