An SEO campaign starts working. Rankings climb, sessions climb, and then the monthly report lands with a number nobody wants to explain: the conversion rate is down. Traffic up thirty percent, conversion rate down from 3.1% to 2.4%. Somebody asks whether the new traffic is junk.
Usually it is not junk. It is the arithmetic of a site that has started ranking for more than its own brand name.
Why does conversion rate fall when rankings improve?
Because site-wide conversion rate is a weighted average, and successful SEO changes the weights. A site that ranks only for its brand name and a handful of transactional queries converts well, because almost everyone arriving already knows what they want. Rank for the broader research queries in your category and you add visitors who are two weeks from a decision rather than two minutes.
Those visitors convert at a lower rate. They also did not exist as visitors last month. The average falls; the total rises. Both are true at once, and only one of them is on the slide.
Conversion rate is a ratio, and SEO changes the denominator faster than the numerator. Judging search work by a site-wide ratio punishes exactly the growth you paid for.
How do you tell dilution apart from a real problem?
Segment before you diagnose. Dilution and breakage look identical in the site-wide number and completely different one level down.
The test is whether conversions fell in the segments that were already converting. Pull organic search sessions split by landing page group — brand pages, service pages, blog — and compare each group to itself over the same period.
| What you see | Dilution | Real problem |
|---|---|---|
| Total conversions | Flat or up | Down |
| Service-page conversion rate | Roughly unchanged | Down |
| Where the new sessions land | Blog and guides | Same pages as before |
| Bounce rate on money pages | Unchanged | Up |
If the money pages convert the way they always did and the extra sessions are landing on informational content, nothing broke. If the money pages themselves are converting worse, the traffic mix is not your explanation and you should stop looking at it.
Which number should replace conversion rate in the report?
Conversions per month, segmented by intent, with the ratio kept as a secondary diagnostic rather than the headline. The ratio answers "how efficient is this page", which is a page-level question. It cannot answer "is search working", which is a business question.
For most service businesses the useful set is small:
- Total qualified conversions from organic, month over month.
- Conversion rate within each landing page group, so dilution cannot hide a regression.
- Assisted conversions, since research-stage visitors frequently return by another channel.
- Revenue or booked value where you can join it, because lead counts hide quality changes.
The trap: optimising the ratio by shrinking the audience
There is a fast way to fix a falling conversion rate. Stop ranking for research queries. Delete the guides, narrow the keyword set, and the ratio recovers within a quarter.
Businesses do this by accident. A ratio on a dashboard creates pressure, somebody notices the blog converts at 0.4%, and the blog gets deprioritised. The ratio improves. The pipeline does not, because those visitors were the top of it.
Cutting informational content to protect a conversion-rate metric removes the audience that would have converted later. The number improves because the business got smaller.
What about visitors who never convert on the first visit?
Most of them, in any considered purchase. This is where the ratio misleads hardest, because a session and a customer are not the same unit and the ratio pretends they are.
Two practical adjustments help. First, track a secondary conversion that a research-stage visitor can plausibly complete — a saved quote, a downloaded checklist, a newsletter signup — so the top of the funnel has a measurable outcome instead of being scored against a form it was never going to fill in. Second, look at your data over a window matched to your actual sales cycle. A thirty-day report on a ninety-day decision will always make research traffic look worthless.
How long does the dilution last?
Until the traffic mix stabilises, which for most sites is one to two quarters after the rankings do. The ratio settles at a new baseline that reflects the wider audience, and comparisons after that point are meaningful again.
The mistake is comparing month four against month one and treating the difference as a verdict. You are comparing two different businesses: one that was visible to a narrow slice of its market, and one that is visible to more of it. The second is the one you were trying to build.
Before the next report goes out, decide which number the business will be judged on — and whether the person reading it knows that a falling ratio can be the receipt for work that succeeded.
Not sure whether your conversion rate is falling or just diluting?
We will segment your organic traffic by intent and show you which queries are earning revenue and which are only earning sessions.
Get in Touch →Does this apply to your site?
Reading about it is one thing. Point the scan at your own site and see whether this applies to you, and what it is worth fixing.
Free and unlimited. No account, no card, and you get every finding rather than a teaser.