AI visibility for cleaning companies: commercial and residential apart

When does an answer count as naming you for the job?

Three verdicts can be read from any answer: named, recommended, first. Our methodology records two of them, named and first. What changes for a cleaning company is what each one is read against. Every line in the panel carries a scope, meaning a service, a building, a schedule and an area, and a verdict means something only if the answer named you for that scope.

• Named: your company appears anywhere in the answer, under any form of its name. If you trade under a franchise brand, decide before the first reading whether the brand without your location counts, and write the rule into the panel file.
• Recommended: the answer puts you forward for the job in the line. Being named as a house cleaner in reply to a question about a janitorial contract, or listed among directories, counts as named, not as an AI recommendation.
• First: yours is the first cleaning company the reader meets. A directory or a booking marketplace listed above you does not change this verdict; it belongs in the cited-page column below.

The scope settles the awkward cases. An answer that recommends you for disinfection after you stopped offering it is not a success to score; it is a wrong fact to trace. An answer that names you for office cleaning with a schedule you do not work is named, with the mismatch written down beside the verdict.

Why office questions and home questions are read as separate panels

The two buyers ask about different things. A facility or property manager's question carries a building type and size, a frequency, working hours, insurance and a monthly contract. A household's carries a date, a home size, a flat price, supplies and sometimes pets. Each side has its own competitors as well: janitorial contractors on one, house cleaners and booking marketplaces on the other.

One number for both would hide the thing worth knowing. A company named for home cleaning in every run and never for office contracts would score somewhere in the middle, and the middle points at no page. Two scores, one per panel, point at the half of the site that gives the answers nothing to work with. The visibility score from 0 to 100 is the methodology's, the share of a panel's questions that named the company with no weighting, applied to each panel on its own.

If you do only one kind of work, run only that panel. What each side asks your pages to state, fact by fact, is on the cleaning companies overview; this page is about reading what the answers make of it.

Twelve bracketed lines: six commercial, six residential

Twelve questions is what we run for a paying client. Split them between the two panels by where your revenue comes from, and within each panel change one thing from line to line: the service, the building, the schedule or the place. Fill every bracket from your own work, then leave the wording alone once the first reading is taken.

Lines 6 and 12 carry your name, so read them apart from the other ten, which are asked by people who have not heard of you yet. Weight the split by your work: a janitorial contractor may keep ten commercial lines and two residential, or drop the residential panel entirely. The dated panel file, three runs per assistant and how to retire a line without breaking the series are in setting up prompt monitoring.

The record: scope, schedule and price basis, and the page behind each

The usual record stays: named, recommended, first, who else was named (the start of your share of voice), and the full answer kept word for word. A cleaning company adds four columns that hold the answer's details against its own pages on the day of the run, filled in by the person who reads the answer.

Nothing we have published tells you what the last column will show. Our audit's probe counted the names in 9,240 answers, one Claude question per business, and 99.4% of them were actual businesses while 0.6% were directories or aggregators. It did not record the pages behind those answers and it has no cleaning-only cut, so how often an answer about a cleaning company leans on a marketplace is something only your own record can show.

What the first two readings point at

• Named on residential lines and absent from commercial ones: read your commercial pages for the building types, sizes, schedules and insurance a facility manager asks about. Where those are missing, the commercial lines have nothing of yours to draw on.
• Named for the right service with the wrong schedule or price basis: open the page the answer cited. An old price page, a directory's default or a review that mentions an hourly rate can each put a basis beside your name that your own page no longer states.
• The brand named and your location not: give the location its own page, with its area, phone number and services, and check what the brand's page says about you.
• Cited pages mostly directories and marketplaces: set their listing of you beside your own pages, publish what yours lack, and correct the listing where you can.

Treat the first reading as a baseline, not a grade. It shows which panel and which pages to work on first; it cannot count the contracts that followed an answer, so ask each new client how they found you and keep the replies beside the panel. For a one-off look before committing to a monthly panel, the afternoon's check runs a facility manager's questions once and traces each wrong answer.

For scale, our probe named 14.9% of 9,240 businesses on one Claude question each, and 17.7% among the 5,928 answers that gave a differentiated list: one assistant, one day, a direction rather than a rate, and not a cleaning figure. The businesses it named were, on inspection, ones whose sites had answer-shaped content and a consistent identity.