The swap test: cover the suburb name and read it again

Open two of your suburb pages in two tabs. Pick the two suburbs furthest apart on the map, because those two have the best chance of being genuinely different.

Copy the text of both into a blank document. Now find every mention of the suburb name in each one and replace it with the word here. Heading included.

Read the two side by side. What you notice in the next 60 seconds is the whole test.

Question 1. With the name covered, is any of the rest untrue?

Plainly: is there anything on this page that is only true of that one suburb?

A pass looks like a street, an estate name, a build era, what a job cost, how long you were on site, a fault you keep finding in one pocket of housing. Something that would be wrong if you pasted it onto the page next door.

A fail is a page that stays completely true after the swap. Every sentence survives being moved 8 km up the road.

Watch for the near miss. A drive-time line, a map embed, a “proudly serving” sentence and a list of the neighbouring suburbs all change words without changing the page. They are variables, not facts. A reader skims straight past them and so does a machine looking for something new.

Question 2. Would a reader believe you had been there?

This is a different question from the first one and it catches different pages. A page can be technically unique and still read as though it was written by somebody who has never been to the place.

The test is narrow. Find one sentence you could only have written after doing a job in that suburb. Not a sentence about the suburb. A sentence about the work you did in it. If you can’t find one, you have your answer.

Here is a made-up pair to show the gap. Say an electrician has two versions of the same paragraph:

Our licensed electricians provide fast, reliable service to homes and businesses across the area and surrounding suburbs.

Most of the houses on the high side went up in the late 70s, so we still find ceramic fuses in the boards up there. The last one we swapped over came to $1,180, and we carry the parts on the van because we know what is behind those covers.

The second one isn’t better writing. It is writing that can’t be moved. Nobody can lift it onto another page without it becoming untrue, which is the only property that matters here.

Question 3. Written for a customer, or filled in for a search?

The give-away here isn’t the prose. It’s the shape.

Line the headings of both pages up in a column, suburb names already removed. If the two lists are identical, the pages came off a template, whatever the paragraphs underneath say. Nobody writing for a customer produces the same 6 headings twice by accident.

Headings for a customer come from what people in that suburb actually ring about. That is a different list in a townhouse estate than it is on acreage, and different again where there is no mains gas or the parking is hopeless. If the questions differ, the headings have to differ.

One matching heading is nothing. It’s a page title. It’s 3 or more in the same order that tells you the page was filled in rather than written.

What this test is not

It isn’t a length test. A 1,500-word page can fail all 3 questions, and the extra words are usually the reason it does. Filler is the easiest thing in the world to duplicate.

It isn’t a writing test either. Polished and interchangeable is still interchangeable.

And it isn’t a score from Google. Nobody outside Google has one of those. It is a test of the thing Google’s published policy calls scaled content abuse: pages produced at volume to catch a search rather than to answer the person who made it. Google ran a spam update in June 2026 and didn’t say what it targeted. The policy it is enforcing is public either way.

3 things to do with a page that fails

Which one is right depends on a single question: have you actually done work in that suburb?

1. Fold it into your service-area page. One page listing the suburbs you genuinely cover, a short section each, kept accurate. If you have never worked there, this is the honest home for the name.
2. Feed it. If you have worked there, the material exists and it is in your head rather than on your site. It takes about 15 minutes on the phone to get it out — that is what the 15-minute call is for.
3. Delete it. If you were sold 25 location pages and you have real material for 7, removing 18 is the correct move, not a defeat. You are not losing pages. You are losing 18 copies of one page.

The one option that isn’t on the list is rewording. Changing the adjectives on a page that fails question 1 produces a page that fails question 1 with different adjectives.

How many pages do you need to test?

Two. If the two suburbs furthest apart in your radius produce interchangeable pages, everything between them is interchangeable as well, and you already know what the rest of the set looks like.

It works on other people’s sites too, and that is worth 2 minutes before you copy anybody. Take the competitor whose site carries 52 location pages and out-ranks you today. Open 2 of his, cover the names, read them. If they fail the same 3 questions, you are looking at a pattern that happens to be working today and is exposed to the next update, not at a standard to build to. Some thin page sets still rank today. That isn’t proof they’re any good.

Doing it by eye is the fast version. The same test can be run by machine over every page at once, which is what a uniqueness check does, and the advantage of the machine is that it produces a number you can hold somebody to. The advantage of your eyes is that you can do it right now, on a page you paid for, without asking the person who wrote it.

Start with the two furthest apart. Cover the name. Read it again.