What to Tell a Client When ChatGPT Recommends Their Competitor
It usually starts with a screenshot. Your client asked ChatGPT for exactly what they sell, in their own city, and it named a competitor. Maybe two. Now they want to know why, and whether you can fix it.
Here is how to handle that conversation so it ends with a plan instead of a panic.
Step 1: Check what actually happened
One answer from one assistant on one day is a data point, not a verdict. Before you explain anything, re-run the question properly:
- Use the exact phrase the client used, and the phrase their customers would use if it differs.
- Ask with web search on. ChatGPT and Claude answer local questions by searching the web, and those answers are what customers see.
- Ask more than one assistant. In our own tests ChatGPT and Claude often named different companies for the same question.
A DidItIndex scan does this for you: it asks ChatGPT and Claude the client's question with live web search, and lists every site each answer cited. Now you have something more solid than a screenshot.
Step 2: Look at the sources, not just the names
The cited sources tell you why the answer came out the way it did. Sort them into three kinds:
- The competitor's own pages. Usually a page that answers the question directly, often a service page or a page written for that city.
- Third-party pages. Directories, review sites, roundup articles, Reddit threads, industry or utility lists. In a local services test we ran, these made up a large share of the citations.
- Nothing about your client. If none of the cited sources mention them at all, the assistant had nothing to work with.
Step 3: Run a side-by-side comparison
Compare your client against the competitor that keeps getting named, using the same question for both. The point is not to prove the competitor is "better". It is to find the specific, fixable differences: a missing service page, no structured data, a thin location page, AI crawlers blocked by a security plugin.
Step 4: Explain it in one paragraph
Clients do not want a lecture on how language models work. Something like this works:
"ChatGPT does not rank websites the way Google does. When someone asks for a recommendation, it searches the web, reads what it finds, and repeats the businesses it can describe most clearly. Right now it found more about [competitor] than about you: a page about exactly this service in [city], plain facts like their warranty and hours, and listings on [directories it cited]. None of that is permanent. Here is what we will change, in order."
That framing is accurate, it does not blame anyone, and it leads straight into the plan.
Step 5: Give them a short, ordered plan
Three to five items, in this order:
- Remove anything blocking AI crawlers. If this is the problem, it is the whole problem.
- Answer the question on the site. A page for the exact service and place, with plain facts an assistant can repeat.
- Close the gap on the sources the assistants used. Complete the directory and review profiles they cited, and earn genuine mentions where the conversation happens.
- Add structured data so machines can read the business type, area, phone and hours without guessing.
- Re-test monthly on the same questions, and compare against the baseline.
What not to promise
Do not promise a placement or a date. AI answers vary between runs and shift as the web changes, and nobody controls them. Promise the work and the measurement. Clients accept uncertainty when they can see a number moving in the right direction every month.
For the full engagement, from scoping to monthly reporting, see how to audit a client's AI visibility. For what a real set of AI answers looks like, see our Katy contractor test.
Frequently asked questions
Why does ChatGPT recommend a competitor instead of my client?
Usually because, for that question, the competitor is easier to find and quote: it has a page that answers the question directly, plain facts the assistant can repeat, and mentions on directories, review sites and forums the assistant searched. Occasionally the client's site is simply blocked from AI crawlers.
Is one ChatGPT answer enough to judge AI visibility?
No. Answers vary between runs and between assistants. Test the same question in more than one assistant, with web search on, and track the same set of questions over time before drawing conclusions.
How long does it take to change what ChatGPT recommends?
There is no fixed timeline. Blocking issues can be fixed in a day, and new pages can be found within days of being indexed, but listings, reviews and mentions build over weeks and months. Measure monthly.
How do I show a client which competitors AI recommended?
Run the same question your client's customers would ask with web search on, and save the full answer with the sites it cited. A DidItIndex report does this for ChatGPT and Claude and lists each company named and each cited site, so you can walk the client through it.
Does ChatGPT recommend you?
Ask ChatGPT and Claude the question your customers ask, live. Free, no account needed.
Run the Free AI Check