Get named when buyers ask AI for a cleaning company.

The questions buyers really ask

Buyers don't type “cleaning company Miami” into AI. They ask a full question, and every detail in it is a condition your company has to meet.

Which janitorial companies in Miami are licensed, insured and bonded, and can send a certificate of insurance before we sign a contract?

The conditions in it:

• licensed, insured and bonded
• a certificate of insurance
• before the contract
• in Miami

The conditions buyers add

• The building: office, medical office, school, warehouse, church
• Size and how often: square feet, people, visits per week
• When: nights, weekends, after staff leave
• Trust: background checks, insurance, a bond, a supervisor to call
• Money and terms: price, what's included, contract length, how to cancel
• Where and what else: the city, and extra services like floors or windows
• Switching: “we're unhappy with our cleaning company”

Collect your real questions

Use the questions your buyers already asked you. Aim for 15.

• Sales calls and walkthrough notes: what did the buyer ask before they said yes or no?
• Emails and quote-form messages: copy the question word for word.
• Bid documents you've answered: every requirement is a condition a buyer cares about.
• Questions your crew hears on site.

Two rules

• A question that didn't come from a real buyer is a “seed”. Mark it, and write where it came from.
• Group your questions into 5 to 8 clusters: price, schedule, building type, trust, switching.

Map each question to a page

For each question, write down the one page on your site that should answer it. If there isn't one, write “missing”.

• One strong page can answer many questions. Don't build a page for every mix of conditions.
• Answer in the first lines of the page, in plain words. Then give the details.
• The swap test: put another city's name into one of your city pages. If it still reads true, it isn't a real city page. Add real local details, or don't publish it.

To write answers AI can quote, use this prompt for your most common questions:

List the 8 questions facility managers ask most before hiring a company for [SERVICE] in [CITY]: cost, frequency, insurance and bonding, background checks, supplies and green products, after-hours access, contract terms, and what happens when something gets missed. Answer each for my company in 40–60 words: the direct answer in the first sentence, then one specific detail.

Use only my facts below. Where a fact is missing, write [ADD: what’s needed] instead of inventing it.

MY FACTS: [PASTE: pricing approach, insurance, vetting, contract terms, guarantees]

Test it the honest way

Ask each question and write down exactly what you see. Don't guess, and don't trust a single answer.

1. Log out, or use a private window. Start a new chat for every question.
2. Ask the question exactly as you wrote it.
3. Write down the date, whether your company was named, and how many other companies were named.
4. If you're named, ask the same question twice more. Count it only if you're named at least 2 times out of 3. AI answers change by person, place and day.
5. Retest the same questions once a month. Keep the same list for 3 months, or the numbers won't compare.

What we ran into: ChatGPT and Gemini answered without an account (Gemini uses a smaller model when you're logged out). Perplexity asked us to sign up. Google asked us to prove we weren't a robot after a few searches in a row, so ask Google's AI Mode by hand, a few questions at a time.

Score it and fix five pages

Score your pages first, because that part is in your hands. Then fix the five pages that answer the most questions.

• Answered the page says every condition in the question, in plain words.
• Partly the page says some of the conditions.
• Missing there's no page, or the page doesn't answer the question.

Then

• Your headline number is “Named in X of N buyer questions”, with the date.
• Lead with your page scores. You control your pages. You don't control what AI says.
• Pick the five pages or sections that would answer the most questions, and fix those first.

A real example: JPSK, Miami-Dade

We ran this whole lesson on JPSK Commercial Cleaning, the example company used in this course, in Miami-Dade.

Tested October 21, 2026. JPSK is the example company in this course. It has no sales calls or emails yet, so every question here is a seed, written from a real source.

What it shows

• Good pages aren't enough on their own. JPSK fully answers 3 of the 12 questions and still wasn't named. A web search found no listings, reviews or mentions of JPSK on any other site. AI names the companies other websites vouch for. Course 07 shows where to get those mentions.
• If you don't publish prices, AI quotes someone else's. Asked about cost, ChatGPT quoted other websites' pricing guides and other cleaning companies' advertised prices. JPSK has no pricing page.
• City pages that are copies don't help. JPSK's 12 city pages are 96% to 98% the same once the city names are swapped.
• AI asks for the conditions. ChatGPT asked back about the type of building, its size, how often, and what matters most.

The map

Swipe sideways to see every column.

One run per question in each tool, logged out, in a new chat each time. ChatGPT with web search on; Gemini on the smaller model it uses without an account. Google AI Mode was not tested (Google asked for a robot check) and Perplexity needs an account. Other companies are counted, not named.

JPSK's five fixes

1. A pricing page. Typical monthly ranges by office size and visits per week, and what's included. Wins: Price and scope.
2. Real local detail on each city page, or fewer city pages. Buildings you clean there, local rules, travel times. Today JPSK's 12 city pages are 96% to 98% the same once the city names are swapped. Wins: Building type (Doral).
3. Say “nights” and “weekends” on the after-hours page. Your real time windows, and who has the keys and alarm codes. Wins: Schedule.
4. A “when something is missed” section. Who to call, how fast you respond, and a named local supervisor. Wins: Trust.
5. A “how often should we clean?” guide and a “switching to us” section. Visits per week by office size; how a handover from another company works. Wins: Schedule, Switching.

Open the full example (Excel file, no email needed).

Let your AI do it

Paste this prompt into ChatGPT, Claude or Gemini with your questions and your page list. It drafts your map. You still run the tests yourself.

Help me build a Buyer Question Map for [COMPANY], a commercial cleaning company in [CITY].

Here are real questions from my sales calls, emails and quote forms, word for word: [PASTE YOUR QUESTIONS]

Here are the pages on my website: [PASTE YOUR PAGE LIST OR URLS]

1. Group the questions into 5 to 8 clusters.
2. For each question, list the conditions in it: building type, size, how often, nights or weekends, background checks, insurance, contract terms, price, city, extra services, switching companies.
3. For each question, pick the one page on my site that should answer it. Write “missing” if none fits.
4. Don't invent questions. If you suggest one, mark it “seed” and say why.
5. Don't suggest a page for every mix of conditions. Name the 5 pages or sections that would answer the most questions, and the conditions each one must cover.

Give me a table I can paste into a spreadsheet: cluster, question, real or seed, conditions, page, notes.

The template has the map with drop-down menus, a scorecard that counts “Named in X of N” for you, and a tab for the monthly retest. Get the Buyer Question Map template