AI visibility checker: does ChatGPT name your business?

The check builds a question asking for the best businesses in your chosen trade and area. Your business name is used to inspect the answer, rather than being included in that recommendation question. That distinction matters: asking a model directly about your company is different from asking it to suggest providers without naming you. Keep the displayed question with the result when you discuss it with someone else.

The tool calls provider APIs and checks the returned text for your business name. It is not a recording of a customer's signed-in ChatGPT, Claude, Perplexity or Gemini conversation. A consumer app can use different models, settings, search features and conversation history. Treat the API result as this tool's sample, rather than a promise that opening the consumer app will reproduce the same answer.

If the answer includes a numbered list, the displayed place refers to that list. It is not a Google position or a universal AI ranking. If a provider is unavailable or the request fails, keep that engine separate from a successful answer that did not name you. An error supplies no evidence about whether the provider would otherwise recommend your business.

'Named' means the returned text matched the business name under the tool's matching rules. Read the excerpt as well as the badge. Check that it refers to your business, especially if another company has a similar name. The mention does not verify the claims around it, and it does not tell you whether a customer followed a link or made contact.

'Not named' means that the business was not detected in that answer. It does not prove that the model has never heard of you, that your website is blocked or that another wording would return the same list. 'Not checked' means the engine did not provide a usable check in that run. Do not turn a missing measurement into a negative result when you compare providers.

Keep the trade, area, question and date together. If you change the trade description or the place, you have changed the question. That can be a useful experiment, but it is not a direct before-and-after comparison. For a repeat check, start with the same inputs and keep the earlier result even when the newer answer looks better.

Open your own website and public business profile. Can someone identify the business, the services offered and the places covered without guessing? Check the telephone number, website address and business name where you publish them. An old trading name or an unclear service description is worth correcting for customers regardless of whether it explains this particular AI answer.

Google's guidance for AI features says the existing SEO foundations still apply; there is no extra technical requirement or special schema needed to appear in those features. That is guidance about Google's Search features, not a claim that all AI providers work identically. OpenAI's ChatGPT Search documentation describes its own search experience. Read the source for the product you are investigating rather than assuming one provider's rules apply to another.

Be cautious about an offer that guarantees a place in ChatGPT from a single edit. Ask what was measured, which product and question were used, and what the proposed work would change. A useful repair can have a concrete acceptance check, such as making a service page accessible and correcting its business details. A universal recommendation promise has no equivalent evidence in this tool's result.

Imagine a joinery firm that serves commercial clients but describes itself only as 'bespoke solutions' on its opening page. This is an example of a useful investigation, not a measured customer outcome. The tool returns a usable answer that does not name the business. The owner saves the question and checks whether the website clearly describes the commercial joinery work and service area.

The first brief could ask for a plain service description, relevant project details that the firm can substantiate and an enquiry route for a site visit or project scope. The firm checks that any photographs have permission and that the contact details are current. Those changes help a reader understand the offer. They are not evidence that an AI recommendation will follow.

After publication, the owner keeps the same trade and area for a repeat check. If the name appears, they record the answer and date. They still cannot attribute that change to a particular sentence without further evidence. If the name remains absent, they have an accurate service page to build on and a record of what was tried, rather than a guarantee that failed silently.

A mention is only useful commercially if an interested person can continue. Follow the business link where the answer supplies one. Check that it reaches the right site, that the service is easy to identify and that the customer can call or send a suitable enquiry. If the answer gives the wrong details, preserve the excerpt and investigate the public sources before requesting a correction through the relevant product.

Choose a next step that fits the work. A maintenance query may lead to a service enquiry, while a larger fabrication job may need drawings and a project location. Explain what the visitor should send and what happens next. Do not make a customer work out how to contact the office after reading a useful answer about the business.

Save the inputs, date, provider status and returned excerpt. Mark whether you inspected the mention manually and whether the details were correct. Add a note about any website or profile change made between checks. This keeps an answer sample, a public-data correction and an actual enquiry as separate events. None should be reported as the other.

Run the check above to create your first sample. If the result raises a question, request a call with the business name, trade, area and the part you want help understanding. The contact page collects your details so we can reply and agree a time. It does not book an appointment automatically or require you to sign up for monitoring before discussing the result.