What I was actually doing
This started as a spreadsheet. We were rebuilding the site and I wanted honest numbers for a results page, so I pulled a year of Search Console data for every client we have a property for and compared it against the year before.
Twelve of those twenty had come to us already paying somebody else for SEO. I knew that going in. What I did not know - and I want to be straight that I did not know it, because I had never lined them up side by side before - was that those twelve sites were all broken in the same way.
The same five things, twelve times
Twelve businesses, twelve different industries, no connection to each other. And close enough to identical that it stopped reading as coincidence:
- An SEO plugin was installed. That is where it stopped
- The plugin applied a default title template - often the only change on the entire site
- Schema was never filled out
- Pages were never optimised even to what the plugin's own analyser was flagging
- The services list did not match what the business actually sold
None of that is exotic. It is the baseline, the part that comes before anything clever. Every one of those sites had a tool installed that was actively telling somebody what was wrong, and nobody was reading it.
What I take from that is not that these providers were lazy. It is that installing the tool had been mistaken for doing the work - which is an easy mistake to make when the tool presents itself as the solution.
What changed, and what we were allowed to change
Content, header structure and schema filled out properly. The services list cleaned up to match the business. In most cases no new pages at all - we were brought in for SEO, not web development, so the site stayed as it was.
That constraint turned out to matter more than anything else on this page: nineteen of these twenty results came from sites we did not build and could not restructure. I had not framed it that way to myself until the numbers were in front of me.
The numbers
Twelve months to 15 August 2026, against the twelve months before it. Every figure is non-branded - searches containing the business's own name were stripped out before anything was calculated, so none of this is people who already knew the company.
Read the counts carefully, because they differ. Twenty accounts are described on this page; the arithmetic below covers the nineteen with measured figures. The twentieth is missing because that client's Search Console access is broken at the time of writing, and I would rather say so than quietly leave it out of the total.
- Came to us already paying someone - 12 accounts, 11 with measured figures: non-branded clicks 619 to 2,050 (+231%), keywords in the top 10 from 3,775 to 10,825 (+187%)
- Never had SEO before - 7 accounts, all 7 measured: non-branded clicks 1,085 to 2,648 (+144%), keywords in the top 10 from 4,055 to 16,521 (+307%)
- The one site we built ourselves - 1 account: non-branded clicks 780 to 1,750 (+124%), average position 57.9 to 19.8, keywords in the top 10 from 487 to 3,115 (+540%)
The comparison is the part I keep coming back to. Sites arriving off a paid provider gained more clicks than sites that had never had SEO at all - 231% against 144% - because more had been left on the table. Sites that had never had SEO gained far more ranking footprint, which is worth sitting with rather than skipping past. Eleven accounts against seven is a pattern, not a proof, and I would rather say that than dress it up.
The single account we built from the ground up moved thirty-eight places of average position. That is the ceiling, and it is what full control buys. The twelve are what is achievable without it.
The question I cannot answer yet
Here is the thing I did not expect, and I want to flag clearly that it is a question rather than a conclusion.
Every one of those twelve sites had a plugin auto-applying a title template. That means every page on the site got its title built by the same formula, usually ending in the same brand suffix. Nobody wrote them; a setting did.
Keyword cannibalisation is what happens when two of your own pages compete for the same search - Google has to choose, it often picks the weaker one, and both underperform. And it made me wonder whether a title template applied blindly across a site is not just failing to prevent that, but actively creating it. If a formula produces near-identical titles across a dozen pages, has the plugin manufactured the overlap it was installed to fix?
There is a reason to take the idea seriously: Google has openly rewritten page titles it judges unhelpful since 2021, so it is plainly making quality judgements about them. Whether it also reads formula-generated sameness as dilution, I genuinely do not know.
Which leads somewhere uncomfortable. If that is what is happening, a badly configured SEO plugin might leave a site worse off than no plugin at all - and "we installed an SEO plugin" would be worse than nothing rather than merely useless. I want to be careful here: that is a hypothesis drawn from twelve sites, not a finding. I do not have the evidence yet.
I am going to find out. Sitewide cannibalisation analysis is the one thing no off-the-shelf WordPress plugin does, and it is the reason we built our own tooling. The accounts and the data are already there. When I have an answer, it will replace this section - including if the answer is that I was wrong.
Four things you can check yourself
You do not need me for any of this, and none of it obliges you to confront anybody:
- Open your SEO plugin and look at the analysis it is already giving you. Then ask your provider to walk you through why those items are still red
- View any page and look at its headings. If everything is the same size, there is no structure for a search engine to read
- Search your business on Google and look for the extras - star ratings, FAQs, opening hours. If they never appear, your schema is probably empty
- Read your page titles in a row. If they are all the same shape with your business name stuck on the end, nobody wrote them
If those four look right, your provider is doing the work. That is a real outcome and it is the one I hope you get.
How this page is put together
Every client here is anonymous - industry only, region at widest, no names and no logos. That is not modesty. Publishing a client's name beside a story about their previous provider is not mine to do, and it would identify the provider too.
The search figures are measured, pulled from Google Search Console and stripped of branded queries before calculation. What I found at takeover is a different kind of claim - it is my direct account of accounts I ran, not a re-audit performed for this page. And the title-template section above is a hypothesis, labelled as one. I am separating those three because a page arguing that other people overstate their work had better not overstate its own.
If you want me to look
Send your web address and we will run our real tools on it and send you what we find. We do not contact your provider, we do not publish it or name you, and the report is yours to do nothing with if that is what you decide.
Common questions
Are you saying my SEO company is not doing anything?
No. Plenty of providers do good work, and often the answer is that yours is one of them. What I can tell you is what I found on twelve sites that arrived here on somebody else's watch, and how to check your own. If your four checks come back clean, you have your answer and you did not need me for it.
Can a badly set up SEO plugin actually hurt my site?
I think it might, and I am not going to pretend I have proved it. Every one of the twelve had a plugin generating titles from a template, and it made me wonder whether that creates the keyword cannibalisation it was meant to prevent. Google has rewritten titles it judges unhelpful since 2021, so it is clearly assessing them. Whether it treats formula-generated sameness as dilution is the part I do not know yet.
Will you contact my provider?
Never - not to verify anything, not for any reason. We do not publish the audit, name you, or use it as an example.
Why is everyone on this page anonymous?
Because naming a business next to a story about the provider they used to pay would identify both of them. The figures are real and measured; the identities are nobody's business but theirs.
What is keyword cannibalisation?
Two or more of your own pages competing for the same search. Google has to choose, it often chooses the weaker one, and both pages underperform. It is invisible from inside a single page, which is why plugins that check pages one at a time never catch it.
My provider says it takes time. Is that true?
It genuinely does - real movement usually takes months, not weeks. But "it takes time" describes work that is happening. The four checks above tell you whether it is. Schema left empty and a plugin analyser full of red flags are not slow progress, they are absent progress.
What does an audit cost?
Nothing, and there is no catch. We show you what we found. If you want help fixing it, good. If not, keep the report.