Method

How this index is built

159 171 businesses across 13 markets, 39 635 websites checked, most recently on 6 September 2026. Everything below is what stands behind those numbers.

Where the businesses come from

Locations are ingested from OpenStreetMap, one country and one kind of business at a time, using the map's own tags: a car dealership is shop=car, a workshop is shop=car_repair, a tyre centre is shop=tyres, and so on. OpenStreetMap data is licensed ODbL and is credited on every page of this site.

This has a consequence worth stating plainly: the index is as complete as the map is. A dealership nobody has added to OpenStreetMap is not here. Coverage is excellent in some markets and patchy in others, and a market's business count should be read as "what the map records", not "what exists".

A location is matched to the nearest indexed town within 45 km. Beyond that it belongs to no town page rather than to a misleading one — a workshop 90 km from a city is not a workshop in that city.

What the website check establishes

Where a business publishes a website, we fetch its home page and, where it links to one, the page a visitor would use to browse stock or book. From that we record whether the site does each of a fixed list of things: shows vehicles, shows prices, takes a service booking, takes a test drive request, values a part-exchange, calculates finance, runs a live chat, has electric-vehicle content, offers any enquiry form at all.

Every one of these is objective and individually verifiable: open the site and look. Each is stamped with the date it was established, because a website can change the day after we check it. Nothing is inferred from the business's size, brand or reputation.

What we deliberately do not do

  • No score, grade or ranking of businesses. These are findings about a website on a given day, not a verdict on a company. A dealership with a plain site can be an excellent dealership, and we have no basis to say otherwise.
  • No ratings of our own. Where a rating appears it was collected by another platform, is labelled as such, and is never mixed into anything we publish. We do not mark those ratings up as review data in our page metadata, because they are not ours to present that way.
  • No claims from review text. Aggregators routinely publish things like "31% of reviewers mention financing". The underlying platforms expose a handful of review bodies against thousands of ratings, so that sentence describes the sample and pretends to describe the business. We do not make it.
  • No inferred ownership. A group here means locations sharing one registered website domain. That is observable. Corporate ownership is not, and we do not assert it.

How vendors are detected

Tools running on a site are identified from the page source, and only where the vendor's own code names itself — a script loaded from a host that belongs to exactly one company, or a marker that company writes. Anything looser produces false positives, and a false positive here puts a wrong claim on a public page.

The consequence: detection is a floor, not a census. A site running a tool we have no fingerprint for is counted as running nothing. Percentages of vendor use should be read as "at least this many", never as "exactly this many".

There is a second, sharper limit. We read the HTML the server sends and do not run the page's JavaScript, so a widget injected later by a tag manager is invisible to us. That is a common way to deploy a chat tool. Where a market shows a very low figure for something, the honest reading is "we did not find it in the page source", not "it is not there".

We also distinguish a chat widget from a messaging link. A WhatsApp or Messenger link opens someone else's app; a widget answers on the page. They are different things to buy and different things for a customer to use, so they are counted separately and never added together.

How brands are attached

A marque is attached to a location when the location's own OpenStreetMap record carries it as a brand tag, when it appears in the trading name, or when the business's own website names it. Each attachment records which of the three it was. We never infer a franchise from geography or from what a competitor nearby sells.

Which pages exist

A town page is only created where the index holds at least three businesses of that kind in that town. Below three there is nothing to compare, and a directory full of pages listing one business is a well-known way to make a site worthless to everyone including its own readers. Pages that fall below the threshold point search engines at the market page instead.

Correcting something

Location, address, phone and website come from OpenStreetMap and can be corrected there by anyone, including the business itself; the correction reaches this index on the next ingest. Website findings re-check periodically and update on their own. For anything else — a wrong group attribution, a brand we have wrong, a business that should not be listed — write to hello@serviceform.com and we will look at it.

Who publishes this

The index is built and maintained by Serviceform, which sells software to automotive retailers. That is worth knowing when you read it. It is also why the method above is fixed and public: the findings are the same whether or not a business is a customer, no business can pay to change what appears here, and every figure is something you can check yourself by opening the site in question.