On this page
The short version
When AI ignores a client, the instinct is to rewrite everything. Don't. First find out how AI is ignoring them, because there are five different ways, and each has a different fix. Then work in this order:
- Week 1: measure and diagnose. Ask the client's buying questions in all five AI systems, every day, and sort each question by symptom. (Week 1)
- Week 2: fix the pages. Home page, service + town pages, and any page AI reads without naming the client. (Week 2)
- Week 3: fix the sources. Google Business Profile, directories, lists and the regulator. (Week 3)
- Week 4: re-measure and report, on the same questions, with ranges. (Week 4)
The five diagnoses are in the table below.
AI ignores your client, but it still recommends someone
The short answer: when AI ignores a business, it's usually recommending a competitor in its place, not saying nothing.
Three real cases from our tracker, LLM Visibility Lens, from 30 September 2026 onwards:
| Business | Named the business | Named any business |
|---|---|---|
| An industrial software company (9 questions) | 0 of 773 answers | 60% |
| An online mobility store (8 questions) | 28 of 1,016 (2.8%) | 58% |
| Kato Electrical, on his 5 Coquitlam questions | 24 of 656 (3.7%) | 45% |
So "AI ignores my client" really means "AI recommends others for my client's questions". That's good news: the slot exists, and the job is to work out why someone else is in it.
Kato Electrical is our public demo, so his numbers are open; the other businesses stay anonymous. I sell the tracker, so weigh my view. The plan works by hand too.
Before day 1: is the client really ignored?
The short answer: check four things before you tell a client they're invisible.
- Are the questions buying questions? Questions with the client's name in them inflate results, and generic questions ("what is a heat pump") rarely name anyone. Use the questions a new customer would ask.
- Is it more than one check? AI answers change from day to day. Five AI systems agreed on the same morning only about 1 time in 5 (5 AIs, same questions).
- Did you look at each AI separately? One might name the client while another never does (which AI recommends local businesses most).
- Is the name being recognised? Check that AI isn't naming the client under a slightly different spelling.
If the client is still invisible after a week of daily checks across all five, they're ignored. Now find out how.
The five ways AI ignores a business
The short answer: look at what AI does instead of naming the client, and the fix follows.
| What you see | Diagnosis | Fix first | Real example |
|---|---|---|---|
| AI names nobody, or rarely anyone | Unclaimed question | A page that answers it | "Panel upgrade cost in Vancouver": a business named in only 17% of answers |
| AI reads the client's site but names others | Cited, not named | Fix that page | A roof replacement page: cited 36 times, named the roofer twice |
| AI names others and rarely reads the client | Missing sources | Directories, lists, regulator | An online store's product pages: cited once in nearly four weeks |
| Named in one town, not another | Town gap | Town page plus local listings | Kato's Vancouver page named him 29 of 30 times; Surrey 0 of 13 |
| Named by one AI, not another | AI-specific gap | That AI's favourite source | Gemini via Google Maps: one business 64%, another 7% |
Most ignored clients have two or three of these at once, so sort each question separately. Background on each: unclaimed questions and spoken questions, cited but not named, missing sources, town gaps and AI-specific gaps.
Week 1: measure and diagnose
The short answer: a week of daily checks, and a diagnosis for every question.
Days 1–2: write 10 to 20 buying questions. Write them in full sentences, the way customers talk: "is there an electrician open right now in Port Coquitlam that can come out tonight", not "electrician Port Coquitlam". Of 50 real buying questions we checked, none had Google search data as written, yet AI named businesses for them every day (questions nobody searches on Google). Our free question finder can draft them from the client's website.
Days 1–7: ask them in all five AI systems, every day. That's ChatGPT, Gemini, Perplexity, Google AI Mode and Google AI Overviews. Record who's named, and every link each answer cites.
Days 6–7: sort every question into a diagnosis from the table above. Note the sources that keep appearing: those directories and lists are your week 3 shopping list.
Week 2: fix the pages
The short answer: make every important page say who the client is, what they do, where, and how to book.
- Home page. Name, services, areas and contact in the first screen. AI rarely cites home pages for buying questions, but when it did, it named the business 31 times out of 31 (which pages AI reads).
- Service + town pages. One for each main service in each town the client genuinely serves. These drew 395 of 695 own-site citations for an electrician and a roofer.
- Every "cited, not named" page. Add the business name, areas and a booking line near the top. The roof replacement page that was cited 36 times and named the roofer twice had none of these.
- Unclaimed questions. Write the page that answers the question directly, in the customer's words, and make the client the obvious answer.
- FAQs and technical basics. Answer real situations ("tonight?", "same day?"), check robots.txt against OpenAI's crawler documentation, submit the sitemap in Bing Webmaster Tools, and add local business structured data.
Don't build town pages for towns the client barely serves. Kato's Surrey page was cited 13 times and named him in none.
Week 3: fix the sources
The short answer: get the client onto the sites AI actually read in week 1, starting with whatever each AI system leans on.
- Google Business Profile. Gemini relies on Google Maps more than anything else, and profile strength decides whether the client gets named from it. Work through Google Business Profile completely: services, areas, hours, photos.
- The directories from your week 1 list. They differ by trade and country: HomeStars for a Canadian electrician, Angi, BBB and Yelp for an American roofer (the directories AI reads).
- The regulator or association listing. ChatGPT linked an electrical safety regulator in 58 answers. Make sure the client's registration matches their business name exactly.
- The "best of" list sites AI cites. One named the electrician in 53% of the answers that cited it. Ask how businesses are selected.
- The same details everywhere. Name, address, phone, services and areas. Google's guide to business details in Search works as a checklist.
Some listings take days or weeks to be approved and picked up. Start them early in the week.
Week 4: re-measure and report
The short answer: same questions, same AI systems, rates with ranges, and next month's priorities.
- Re-measure the same questions in all five AI systems. Never add or swap questions mid-plan, or the numbers stop being comparable.
- Report rates with ranges, not screenshots. The format we use is in how to report AI visibility to clients.
- Look at each AI system and each town separately. One overall number can hide a big win in one place and nothing in another (why a single score can mislead).
- Pick next month's three gaps, usually the weakest town or the most-cited page that still doesn't name the client.
What 30 days can and can't do
The short answer: 30 days is enough to diagnose and fix; it's often not enough to prove the fix worked on a single question.
- A single question is noisy. A week of daily checks in five AI systems gives about 35 answers per question, so a change of 10 or 15 points can still be chance.
- Groups of questions are clearer. Ten questions over a week is about 350 answers, so the range shrinks to a few points.
- Some fixes show quickly, some don't. A page the AI already reads can change its answers fast. A new directory listing has to be approved and then found, which can take weeks.
So treat day 30 as the first checkpoint, not the finish line, and keep tracking the same questions. When we have before-and-after numbers from running this plan, I'll publish them.
A script for the client
"AI isn't ignoring your market. It's recommending someone else for your customers' questions. In the first week we'll find out exactly why, question by question. Weeks two and three fix the pages and listings AI relies on. In week four we measure again, the same way, and show you what moved and what needs longer."
How we run it
LLM Visibility Lens asks the client's questions in all five AI systems every day. It shows who gets named, which pages and directories AI read, and an action plan for the gaps, and it builds the monthly report under your agency's name. See it on Kato's account in the live demo, compare plans (they start free), or see the agency plan. If you'd rather do it by hand, start from the printable AI visibility checklist.
More reading:
- How to get recommended by ChatGPT
- Which websites ChatGPT trusts
- Invisible on Google, recommended by AI
- 5 AI systems, 35 questions
- Google AI Mode for local businesses
- How to rank in AI Overviews
- AI Overview tracking for near me searches
- ChatGPT citations for local businesses
- How to rank in ChatGPT
- Why there's no rank #1 in ChatGPT
- What one free check can and can't do
- We asked AI who to hire 80 times
The method is on the how it works page. There's also a comparison with Semrush's AI visibility tool, the changelog, the FAQ page and the security page. If you'd like me to diagnose a client with you, get in touch.
Questions people ask
Why does AI ignore my client?
Usually for one of five reasons: nobody owns the question yet, AI reads the client's page but names others, the client is missing from the directories and lists AI reads, they're only known in some towns, or one AI system relies on a source where they're weak. Look at what AI does instead of naming the client to tell which.
What should I fix first when ChatGPT doesn't recommend a client?
Measure for a week first, then fix the client's own pages: the home page, service + town pages, and any page AI cites without naming them. Then work on the sources ChatGPT reads, such as list sites, BBB and the relevant regulator or association.
How long does it take to get a client into AI answers?
Some changes show within days when AI already reads the page you improved. New directory listings can take weeks to be approved and picked up. Plan for a 30-day first checkpoint and keep tracking the same questions after it.
How do I prove to a client that the plan worked?
Track the same questions in the same AI systems before and after, and report the share of answers naming the client with a range. Use groups of questions rather than single ones, because one question gives too few answers to show a reliable change in a month.
Should I create new pages for every town?
Only for towns the client genuinely serves. In our data, an electrician's Vancouver page named him in 29 of 30 answers that cited it, while his Surrey page, for a town where little else connected him, named him in none of 13.
Can I run this plan without a tracking tool?
Yes, by asking the questions yourself each day and recording the answers in a spreadsheet. It's slower, and it's hard to keep five AI systems consistent by hand, which is why we built a tracker, but the plan itself doesn't depend on one.
Harjeet Singh Bedi Founder, LLM Visibility Lens · SEO consultant, Bedi SEO
Harjeet has worked in SEO since 2010, first in agencies and now running Bedi SEO for local businesses, online stores and agencies. He built LLM Visibility Lens after clients kept asking whether they showed up in ChatGPT and other AI tools and nobody could give them an honest, measured answer. Every figure in this article comes from answers his tool collected.