How often does the AI recommend your landscaping business, town by town?

The route map as a list, and every place your area is written

Start with the route written out as a list: every town you serve, the day you are there, and the towns at the edge you turn down. Add three towns you do not serve: one just past the edge of the route, one across a line you do not cross (a river, a county line, a highway), and one whose name is close to a town on your list. The last is there because a near-miss name is an easy mistake for anyone reading place names quickly.

Then open every place your service area is written down, because each one can be the source of a wrong answer: the service-area text on your site (the contact page, the footer, each service page), areaServed in your structured data (view the page source and search for it), the service area in your Business Profile, each directory profile you know of, and any old page that names a town. That last group covers a former address, a post about a job in a town you have since dropped, and a PDF price list with a delivery zone. A sheet with a row per town and a column per source holds all of it.

Two questions per town, asked the same way of all four assistants

Ask two questions about each town. The first is a service question with the town in it, for a service that is in season on the day: "who does fall cleanups in [town]". The second names you: "does [company] work in [town]?" The first shows whether you are named for a homeowner who does not know you yet. The second shows what the answers say about your area once your name is already in the prompt.

Keep the wording identical from town to town, so that only the town changes. Start a new conversation for each question, so one answer does not lean on the last, and turn web search on wherever the assistant offers the choice. Write the town into the question rather than "near me", so the answer is about the town on your list. Paste each answer into the sheet with the date, the assistant and any sources it showed.

Six towns on the route and three off it make 72 answers, one run each. That fits in an afternoon if each answer is read for two things only: whether you are named, and which towns it puts you in.

Four marks on the map

Every answer gets one of four marks, and the mark says where to look next.

When an answer lists the towns you serve, copy the list into the sheet exactly as it was written. A list that matches one of your sources word for word, the footer, the Business Profile or a directory's category page, points straight at that source.

Tracing a wrong town to the page or listing that states it

Work through the off-route marks first, then the missed towns. For each one, go along the sheet's columns and find the source that states the area the answer gave.

The markup row deserves a closer look, because it is the one source no visitor ever sees. schema.org's areaServed takes a place, an administrative area such as a City, a shape or plain text. Written as a list of towns, it can be checked against the page line by line. Written as a GeoCircle, it covers whatever the radius reaches, towns across the river included, and schema.org reads a geoRadius given without a unit as meters, so a radius typed as a number of miles describes a far smaller circle than the one intended.

For the profile, Google's Business Profile guidelines give a business with no storefront that travels to its customers one service-area profile, and say the service area should not extend farther than about two hours of driving from where the business is based. The guidelines allow a larger area for some businesses, and two hours is a limit rather than a size to fill: the profile should name the towns you drive to, the same ones as the site.

Change your own site first, the pages and then the markup, then the profile, then the directories, and write the date of each change in the sheet so a later reading can be set against it. One service area, stated the same way everywhere, is part of what lets a business be read as one entity. A town on the route that is never named needs to be written where the others are, and if you have done a job there, its project record can name the town in a sentence about real work.

Why the fix is never a page per town

The shortcut that suggests itself, a page for each town with the same text and the town's name swapped in, is named in Google's spam policies. Their examples of doorway abuse include pages targeted at specific regions or cities that funnel users to one page. Google's guide to generative AI search adds that creating separate content for every variation of how people search, primarily to manipulate rankings or AI responses, violates its scaled content abuse policy.

The towns belong in one list, on the service-area page, in areaServed and in the Business Profile, the same in all three; the landscaping and lawn care page sets out what else the site states. Where a town needs more than its name on the site, it gets it from a page about something that happened there. A project record is that kind of page: it would exist whether or not anyone ever searched for the town.

What an afternoon's answers are, and what the monthly panel reads

An afternoon's answers are a snapshot: one run of each question, on one day. Ask the same question again and the list can change. Our own largest reading had the same limit. In the probe attached to our audit of 9,483 small business websites (40 trades, 90 metros, late August to early September 2026), Claude named 14.9% of 9,240 businesses on one buyer question each, and we call that figure directional for exactly this reason: one question, one assistant, one day.

So the afternoon tells you where your area is written wrong and which pages and listings to fix. How often you get an AI recommendation, and whether the fixes moved it, is read by the monthly panel, with each month set against the same month a year earlier. Run the afternoon check again whenever the route changes, because a town added or dropped has to change in every column of the sheet.

If you would rather see a first reading before doing any of this, the free visibility check costs $0: a person runs it and sends a written result within 48 hours.