AI visibility for a business in Manchester: a nationally cited British source also won here, and the footprint is Greater Manchester

Manchester is the one British city we measured where a source the engines already lean on nationally also took the local answer. Across twelve Manchester readings on ChatGPT and Gemini, run on 15 and 16 October 2026, all twelve returned a source list and four sources tied at the top: mainchester.co.uk on 3 readings, popula.co.uk on 3, rank4ai.co.uk on 3 and margen.net on 3. We appear nowhere in those twelve. Two of those four were already in the engines' national source lists. rank4ai.co.uk was cited in 39 of the 204 United Kingdom readings in the market run the day before, and margen.net was among Gemini's most cited domains across its 291 readings, at 10. No London or Birmingham source in this set did that. Note what rank4ai.co.uk is called: a category and a country, not a place. It won at both levels, which points the work at scope rather than at the city name, and scope is exactly where a Manchester business has a decision to make, because almost none of them trades in Manchester alone.

The Manchester row, and the source that won twice

rank4ai.co.uk, in the UK market run of 15 October 2026	39 of 204 UK readings
rank4ai.co.uk, in the Manchester city readings	3 of 12
margen.net, in Gemini's 291 readings	10, and 3 of 12 in Manchester
The other two at the top	mainchester.co.uk 3, popula.co.uk 3
Manchester readings that came back with a source list	12 of 12, the highest of the three British cities measured
Manchester readings naming our own domain	0 of 12
Read on	ChatGPT and Gemini, 15 and 16 October 2026, six buyer questions each

Manchester was one of ten cities in that run, 120 readings in all, and not one of them cited us. The engines behind every count above number two, out of the five the product follows: Perplexity, Claude and Google AI Overviews were not questioned here, which is worth settling before any figure above gets quoted as a fact about AI. How the questions were built and how the counts were classified is published in what AI assistants actually cite across 597 readings, and the British market layer above the cities is in the UK cut of 204 readings.

What crossing over actually means

A source cited 39 times in the national set and again in the city set is doing two jobs: the general British question and the specific Manchester one. An assistant has to decide whether a given page is about Manchester or about Britain, and it decides from the text. One page trying to be both gives it nothing to decide with, so a site aiming at both jobs writes two:

• A British layer. The category question answered for the UK, in pounds, with the VAT position stated and a date on it.
• A place layer. The same question answered for the footprint you genuinely trade in, with the towns and boroughs named, and the detail only somebody working there would know.

Greater Manchester is ten boroughs, and your page probably names one

This is where Manchester differs from the other two British cities we measured. London buyers are mostly buying from a national market, and Birmingham buyers tend to name the surrounding towns. A Manchester business usually trades across Greater Manchester, and very often across the North West: Salford, Trafford, Stockport, Bolton, Bury, Oldham, Rochdale, Tameside and Wigan alongside the city itself, then Warrington, Cheshire and across to Liverpool and Leeds for the ones selling regionally. "Manchester" in a buying question is a regional word wearing a city name.

That gives a page two ways to go wrong, both common:

• Under describing. The site says Manchester, the business covers the whole of Greater Manchester, and every buyer in Stockport reads a page that never mentions Stockport.
• Over claiming. The site says the North West, the business really covers three boroughs, and the page cannot answer a Lancaster question at all.

The fix is a written footprint. List the boroughs and larger towns you actually work in, say where a travel charge starts, and say where you will not go. An area page earns its own URL by carrying something only somebody working in that borough would know; without that it is the city page under a different name. What a package should cost per location, line by line, is in what a local package costs per location and the line items it has to name, a useful sanity check before anybody sells you forty of them.

The multi site answer, which Manchester asks for more than most

Manchester's retail, leisure, hospitality and professional services operators often run several sites across the North, and the question an assistant keeps having to answer is "do they cover where I am". That answer gets assembled from facts, so write them down per site:

The fact	Why an assistant needs it
Trading name and address of each site	Tells the engine these are branches of one business, not several businesses with similar names
What each site does and does not do	Stops a buyer being sent to the branch that cannot help
Opening hours per site, bank holidays included	The most expensive error in local trading, and it usually arrives from a stale third party listing
Prices in pounds with the VAT position, per site if they differ	A commercial buyer in Salford and a consumer in Didsbury are asking two different price questions
Registered company name and number at Companies House	Ties every branch to one entity somebody else can confirm
Coverage written as a borough and town list	Answers the question that was actually asked, which is whether you come to where the buyer is

What the engines said they would cite instead

The build spec conversations ran alongside the readings: each engine was shown its own answer and asked what a page would need to contain to be cited in its place. Three of the counts matter most to a business whose scope is the thing in question.

• 101 of 102 wanted a before and after with the window stated. Not a result on its own: the starting position, the end position, and the dates either side.
• 98 of 102 wanted the engines named rather than described as AI in general, which is why every figure on this page says ChatGPT and Gemini.
• 80 of 102 wanted the sample size. Twelve readings is a small number, printed as twelve, because a reader can weigh twelve and cannot weigh "our research".

One British constraint sits across all of it, and in Manchester it lands on coverage rather than on price. Coverage is an advertising claim like any other, which in this country puts it inside the CAP Code with the ASA able to take a complaint about it. So "we cover the North West" has to be true of the North West. A borough list is the safer version and it is also the more quotable one.

Footprint and scope: the Manchester questions that follow

Which sources did ChatGPT and Gemini use for Manchester?

mainchester.co.uk, popula.co.uk, rank4ai.co.uk and margen.net, each cited in 3 of the 12 readings on 15 and 16 October 2026, with all 12 readings returning a source list. rank4ai.co.uk was also cited in 39 of the 204 UK readings measured on 15 October 2026.

Should we write one Manchester page or one page per borough?

One page for each borough where you have real work to describe, plus a single Greater Manchester page covering the rest. The test is whether the page would survive having the town name swapped out. If it would read just as well, it is not a second page.

Does the city need to be in our domain name?

No. At least two of the four sources that won Manchester carry nothing about the city in the domain at all, and the one that also won nationally, rank4ai.co.uk, names a category rather than a place. Put the footprint in the text, where it can be read and quoted, rather than asking the domain to carry it.

We are based in Manchester but sell nationally. Which page wins the city question?

The one that says which it is in its first paragraph. Write the British page for the national buyer and a Manchester page for the local one, and let each be unambiguous. The source that crossed over in this run was working at both levels.

How would we repeat this measurement ourselves?

Take the six questions your Manchester customers actually ask, in their words, with no business name in them. Put each to ChatGPT and to Gemini, write down who was named and which sources were cited, then repeat it against the same frozen wording each month.

The smallest number in the table is the one to act on

Third place in Manchester was three readings. A field whose leading source appears in three of twelve answers has nobody holding it, and the firms in those slots are small British operations publishing their own pages. A Greater Manchester business that writes down its boroughs, its prices in pounds with the VAT position, its company number and the window it measured over is competing for a slot nobody owns. We sell both halves of that, the measurement and the pages, so the Manchester number we are judged by here is our own: zero citations in twelve readings. AI Knows Us published it rather than leaving it out, because a page arguing that scope should be written down has no business hiding its own.