Six near-identical pages, one got 5,926 impressions and one got 6

We published six industry pages on the same day, from the same template. They have almost the same word count, the same heading structure, the same schema, and the same number of outbound links. Ninety days later, one of them had earned 5,926 search impressions and another had earned 6. That's roughly a thousand to one, from pages a person would struggle to tell apart.

We went looking for the reason and tested four explanations. Internal links, crawl frequency, search volume and page structure. All four failed. This is what the numbers actually show, which explanation is still standing, and the experiment we're running now to test it. We're publishing the version where we were wrong, because the disproof turned out to be more useful than the tidy answer we expected to find.

The short version: six pages, one template, one publication date, a thousand to one spread in impressions. Internal links didn't explain it. Neither did crawl rate, search volume or structure. The only remaining difference is that the page that worked contains a real named client result and the others don't.

What we shipped, and what happened

On 1 July 2026 we published six pages targeting SEO for different industries. Same shared template, same build, same sitemap submission. Here is where they stood after 90 days, alongside the number of internal links pointing at each one from across our site.

Every one of the six is indexed, crawlable, has a matching canonical and passes structured data validation. None of them is broken.

The four explanations that didn't survive

Internal links. This was our first theory and it looked convincing until we counted properly. Electricians and heating engineers carry exactly the same 24 inbound links as plumbers, and they returned 597 and 155 impressions. Accountants has 4 inbound links and beat heating engineers, which has 24. Our first pass compared only two of the six pages and produced a much neater answer than the full data supports.

Crawl frequency. Heating engineers was last crawled on 31 July and has 155 impressions. Accountants was last crawled on 12 July, nineteen days earlier, and has 243. Google is visiting all six.

Search volume. This one contradicts the result outright. According to Semrush, "seo for dentists" is searched around 1,300 times a month in the UK at a keyword difficulty of 9. "Seo for plumbers" is around 880 a month at difficulty 6. The page targeting the bigger, barely harder term is the page with 6 impressions.

Page structure. We compared the two extremes line by line. Dentists runs 1,259 words against 1,229 for plumbers. Both have 8 H2s and 6 H3s. Both carry four blocks of structured data. Dentists has 24 outbound internal links, plumbers has 23. The page that failed is marginally longer and marginally better linked.

The one difference left standing

After all that, one difference remains. The plumbers page names The Boiler Co twelve times: a real client, in that exact trade, with results we can evidence. The other five pages have no named result from their own industry.

We can't prove that's the cause. Six pages is a small sample, we changed nothing deliberately, and the case study also makes that page longer in unique, non-templated ways, so more than one thing varies at once. Correlation, not causation, and we'd rather say so than dress it up.

But it's the only candidate we haven't been able to knock down, and it points somewhere uncomfortable for how industry landing pages usually get built. Spinning up twelve variations of a template is quick. Having a real result in each of those twelve industries is slow, and it might be the part that actually matters.

The experiment we're running now

Rather than argue about it, we're testing it. As of today the dentists page carries a real result from a healthcare and wellness client, and nothing else about it has changed. No new links, no restructure, no rewrite of the meta description. If impressions move on proof alone, that's a reasonable signal. If they don't, our last explanation dies too and we'll publish that as well.

We'll update this page with what happens either way.

The same thing shows up in a client's data

While writing this we went back through twelve months of Search Console data for a UK healthcare and wellness practice we worked with, and found the pattern again on a completely different site.

Their average position improved from 19.2 to 7.3 over the engagement, and twenty-one of their condition pages reached page one. Their monthly impressions went from roughly 8,500 to over 22,000. Clicks rose by about a third, and their click-through rate fell from 1.20% to 0.61%. One of their pages ranks first for its target term, has been served over a thousand times, and has never been clicked.

Two unrelated sites, different industries, different platforms, same shape. Visibility went up a great deal and clicks barely moved. That's worth knowing before you buy a package sold on impressions or average position, because both of those numbers can improve substantially while nothing reaches your inbox.

What to check when a page is indexed and invisible

If you've got a page in this position, the order below is roughly what we ran, and it's cheap to repeat.

1. Confirm it really is indexed. Use URL Inspection rather than assuming. Ours came back clean, which ruled out a month of technical work.
2. Check the target term has volume and reachable difficulty. If the term is a fight you can't win yet, the page isn't the problem.
3. Compare it against a page of yours that did work. Word count, headings, schema, links in and out, lined up in a table.
4. Then check the pages that also failed. This is the step we nearly skipped, and it's the one that killed our theory.
5. Ask what the page says that the current results don't. If the honest answer is nothing, that's your finding.

Steps 1 to 4 take an afternoon. Step 5 is the hard one, and it's usually the answer.

If you're reading your own data

The thing worth taking from this isn't a rule about internal linking. It's how nearly we talked ourselves into one.

Two pages made a clean story. Six pages destroyed it. When one page takes off and you go looking for the reason, the useful question is whether the pages that didn't take off share the feature you're about to credit. Ours did, and that was the end of the theory.

It's the same reason we argue for working out the strategy before buying the tactics. A confident explanation built on the two data points that suit you is worse than no explanation, because you'll go and spend money on it.

If you want a straight read on why your own pages aren't landing, our SEO service starts with exactly this kind of look at the data, and you can see the two pages in question at SEO for dentists and SEO for plumbers.