Every SEO audit tool, ours included, is tempted by the same pitch: fix these errors and you will rank higher. Before saying it, we tested it. If SEO errors affect rankings, the sites at the top of Google should carry fewer of them than the sites a few places down. So we took the first page of Google for 25 everyday searches, ran the same website audit on the sites at positions 1 to 3 and the sites at positions 6 to 10, and checked every claim the audit made by hand.
The answer was no. Duplicate pages, thin pages, reused titles and the rest were about as common at the top of page one as lower down.
How we checked whether SEO errors affect rankings
We wrote the method down before scanning a single site, including what we would measure and how we would compare it, so the result could not be chosen after we saw it. A result of “no difference” was to be published the same way as any other.
- The searches. 25 searches a customer would type: local services in the UK, the US and Australia, online shops, software, how-to questions, directories and three searches in other languages. We do not list them, because the searches would name the sites.
- The sites. From each first page, the first three results that were one owner’s own website, and the last three. Marketplaces, social networks, review platforms and directories of other businesses were set aside, since they are not one owner’s site (in the directory searches, directories were the point and stayed in). Searches left without two sites in either group were dropped by the same rule, which left 21 searches and 123 sites.
- The audit. The same deep scan on every site: duplicate and near-duplicate pages, thin pages, reused and templated titles and descriptions, and the technical faults that come with them.
- The check. Every claim the audit made was checked against the site’s own pages, independently of the audit, and judged right or wrong by hand.
What we found: the top results had as many SEO errors
Within each search we compared the two groups. The lower-ranked sites had the higher audit score in only 7 of 19 searches compared, and across all of them the typical gap ran slightly the other way: the top three scored 6 points more. Over all the sites we could read, the two groups look alike:
| Positions 1 to 3 | Positions 6 to 10 | |
|---|---|---|
| Sites read | 46 | 47 |
| Average audit score (higher means more found) | 46.8 | 46.1 |
| Average findings per site | 3.00 | 3.04 |
| At least one near-duplicate page pair | 70% | 60% |
| A real share of thin pages | 59% | 62% |
| The same title on different pages | 50% | 62% |
| The same description on different pages | 46% | 45% |
| Titles or descriptions differing by a word | 46% | 47% |
| One page served at two addresses | 9% | 9% |
Removing the claims we judged wrong changes none of this: the lower-ranked group still came out higher in 7 of 19.
What this does and does not mean
It means that on page one, these errors do not explain who ranks above whom. A site can rank first with near-duplicate pages and reused titles, and the site in eighth place is no more likely to have them. Whatever separates position two from position nine, it was not visible in this audit.
It does not mean the errors never matter. We compared positions within page one, not page one against page five, and we measured what sites have, not what happens when they fix it. Links, brand, relevance to the search and the rest of what Google weighs were not measured at all, and they are the obvious candidates for what does separate the top from the bottom.
What it rules out is the easy claim. If a tool tells you its score predicts your ranking, or that fixing its list will move you up, it is promising something we could not find on 123 real sites.
Should you still fix what an SEO audit finds?
Yes, for what the problems are, not for a ranking promise. Some findings are plain defects that hurt whoever meets them:
- One page published at two addresses, so links and visitors split between copies.
- Links inside the site that lead to pages that no longer exist.
- A sitemap that lists dead or redirected pages instead of the pages themselves.
- An article showing at the address of a different article.
Others are true but harmless: a product in four colours is four pages that are supposed to read alike, and a series of guides will have titles that differ by a word. A good audit tells you which is which and names the pages, so you can open them and decide.
How accurate was the audit itself?
We checked all 534 claims the audit made on these sites, none of which it had seen before. 29 were wrong, about one in twenty, and 40 were true but described something the site does on purpose. The most common wrong ones were pages the audit misread: a “too many requests” or “verify you are human” page taken for the site’s own content, or a block repeated across one section taken for content the pages share.
One in twenty is good enough to point at the right pages and not good enough to act on unread. That is why every finding we send to a site owner is checked by hand first, and why our reports name the pages behind every number. An audit you cannot check is a verdict you have to trust.
Frequently asked questions
Do SEO errors affect rankings?
Not in a way we could see on the first page of Google. For 21 searches we audited the sites at positions 1 to 3 and the sites at positions 6 to 10 with the same tool. Duplicate pages, thin pages, reused titles and the rest were about as common at the top as lower down, and the lower-ranked sites had the higher score in only 7 of 19 searches compared. That does not prove the errors never matter. It means they do not explain who ranks above whom on page one.
Do top-ranking websites have fewer SEO problems?
No. The sites ranking first to third carried as many audit findings as the sites ranking sixth to tenth: an average of three scored findings each in both groups. Most top results had at least one near-duplicate page pair, and half had titles reused across pages.
Should I still fix the problems an SEO audit finds?
Fix the ones that are real defects, for what they are: one page published at two addresses, links that lead to pages that no longer exist, a sitemap that lists dead or redirected pages, an article showing at the wrong address. Visitors and crawlers run into those whatever they do to rankings. Be wary of any audit that promises a ranking gain for fixing them.
How accurate is an automated SEO audit?
Ours, checked claim by claim against the live pages on sites it had never seen, was wrong about one time in twenty. That is why every finding we send to a site owner is checked by hand first, and why a report should always name the pages behind each claim so you can check it yourself.
Can any tool tell me why my site dropped in Google?
No. Nobody outside Google can see why its systems ranked a page where they did. A tool can show you what is on your own pages and how they relate to each other. Anything that claims to know the reason for a ranking change is guessing.
Related: website mistakes to avoid, from 50 verified websites, the duplicate-page study of 30 sites, and how Google ranks pages.