It is easy to argue about what makes a location page good. It is harder to argue with a number. So here are three location pages for the same trade, a Gas Safe boiler engineer working across Uckfield and Lewes, each scored against a stated rubric so you can see exactly why one would rank and another would get penalised.
The rubric below mirrors the three-part model the Quality Score uses, out of 100 points: sibling similarity (how different this page is from your other location pages), local substance (whether the page says anything true and specific about the place), and a thin-content floor (word count, one H1, a meta description set). The pages are illustrative, but the scoring is the real model, so a reader could open the same page in the plugin and reach the same floor and similarity numbers shown here.
How to read the scores. A page below roughly 60 is the kind of thing Google’s spam policies describe as doorway or scaled-content risk. Sixty is also Townsmith’s own default publish threshold, the score a page has to clear before the plugin will let it go live. That gate is Townsmith’s model, not a number Google publishes, so treat it as a sensible internal bar rather than a guarantee.
Example one: the page that gets you flagged (score: 18)
This is the page most plugins produce by default: the service description with the town name swapped in. “Looking for boiler repair in Uckfield? Our Gas Safe engineers cover Uckfield and the surrounding area. Call today for fast, reliable boiler repair in Uckfield.” It has a hero image and not much else.
- Sibling similarity: 0 / 45. Once the service and area names are stripped out, the leftover wording is 88% identical to the Lewes page, well past the point where the score gives any credit. As far as the model is concerned, this is the same page twice.
- Local substance: 5 / 35. There is a hero image, which earns the single local-extra point, but no filled local content slot and no real area descriptor, so the two substance components that actually reward saying something about Uckfield both score nothing.
- Thin-content floor: 13 / 20. Around 70 words, well short of the 250-word target, so it earns only a fraction of the word points (about three of ten). It does have one H1 and a meta description set, which collect the remaining ten.
Total: 18 out of 100. Far below any sensible threshold. This is the page that earns a manual action, not the page that earns a phone call.
Example two: the page that scrapes a pass (score: 54)
This one has had some real work put into it. The author has written two genuine paragraphs about working in Lewes, mentioned the kinds of older properties common in the town centre, and added a short note about parking and access. It is not lazy. It is just not quite finished.
- Sibling similarity: 22 / 45. Genuinely better than example one, but the opening and closing sections are still lifted almost word for word from the other town pages. The unique middle saves it; the templated top and tail drag it down.
- Local substance: 12 / 35. The local slot is part-filled and the area descriptor is shorter than the target, so each earns partial credit, and there is no second local extra such as an embedded map or an area-specific FAQ.
- Thin-content floor: 20 / 20. Over 250 words, exactly one H1, a meta description set. The floor is the easy part to get right, and this page does.
Total: 54 out of 100, just under a sensible threshold. The frustrating part is how close it is. Half an hour of specific local detail, in the slots and in the descriptor, would clear the gate comfortably.
Example three: the page that earns the work (score: 93)
This page reads like it was written by someone who has actually worked in the area. It names a real job, sensibly anonymised: a Worcester Bosch Greenstar that lost pressure on Framfield Road, the kind of detail a homeowner recognises and a competitor cannot copy. It explains what is different about working on the older housing stock, links to the relevant service page, and carries a photo of an actual install rather than stock art.
- Sibling similarity: 40 / 45. The worst-matching sibling sits at about 61% similarity once names are stripped, comfortably inside the band that scores well. The shared wording is the unavoidable scaffolding; everything that carries meaning is local and different.
- Local substance: 33 / 35. The local content slot is nearly full, the area descriptor clears its word target, and the page carries a genuine local extra, a photo of a real job. Strong content that says something true about the place.
- Thin-content floor: 20 / 20. Well over 250 words, one H1, a meta description set. Full marks, as you would expect from a page this substantial.
Total: 93 out of 100. This is the page that ranks, because it is the page a human would actually want to read. The score is not rewarding length or keywords. It is rewarding the one thing thin location pages never have: something specific and true to say about the place.
What the three pages have in common
The gap between 18 and 93 is not effort spent on the template. All three pages have a heading, an image, and a meta description. The gap is whether anyone wrote anything that only applies to that one town. The plugin scores and advises; it does not write the local detail for you, because the local detail is the part that has to be true. You bring the Framfield Road boiler; the score just tells you whether you remembered to.
If you want to turn one of your own thin pages into the third example, start with the local landing page content guide, add real photos, and check your wording against the duplicate content guide. When you are ready to see the number for yourself, the Quality Score documentation explains every part of the rubric, and the features page shows how it fits the rest of the workflow. The location page template and the guide on measuring location pages cover the structure and the follow-through.