Published October 21, 2026
How to test if AI recommends your website, for free
Choose one question a customer would ask without your brand name, ask the assistants and save the complete answers. Record a positive business name in the response separately from a site link among sources and from an unavailable answer. The sayurname free test sends one question to four assistants and stores the available responses as the starting point for that question; it requires a site and email, not a payment card.
Start with the question, not your brand
Pick a question a customer would ask before knowing you exist, such as "what's the best accounting software for freelancers?" rather than your company name. Searching for your own name only tells you that the AI knows you exist. It does not tell you whether it recommends you.
Include the service and location that matter to the customer, and avoid inserting the answer into the question. A named question can diagnose what an assistant says about your brand, but it answers a different question from whether an unknown customer discovers you. Keep the two questions in separate series.
Test it by hand
In each available assistant, start a fresh conversation and ask the same question. Save the complete answer with the date, assistant and visible search setting. Record whether the business is named positively in the answer, which other businesses appear and which sites are listed as sources. A site linked only as a source is not a positive name in the answer.
Availability and settings can differ between a manual session and an automated test. Record what you can observe and do not imply the two methods will reproduce identical responses. A failure to get an answer is an unavailable result, not evidence that the business was absent.
Why one try is not enough
The same question can return a different answer on another day or in another conversation. A single answer is a snapshot. To see a pattern you need the same wording, a saved copy of each answer and a second look later.
A second run is useful only if you keep the question and relevant conditions comparable. Note any visible model or search change. Different wording, a new location or a prior conversation can alter the question being answered; start a new series when the question changes and retain the first result.
The sayurname free test
You enter your website, one question your customers ask and your email. We ask that question to ChatGPT, Gemini, Perplexity and Claude, and save each full answer as your starting point. It is free and needs no card.
The test covers one question. If you subscribe, you add the others that matter to your business, and each new question starts from its own first measurement.
The test runs one question against ChatGPT, Gemini, Perplexity and Claude; an individual provider may fail, so review what is actually available. The saved free-test result is the starting point for that question. A later question receives its own first measurement. The service has usage limits and a message may explain when a test cannot be started.
Worked example: a clinic tests one question
This example is hypothetical; the clinic is not a sayurname customer and the results are invented. A small physiotherapy clinic in Austin treats runners after injury. It has space for six new appointments each month and wants to see what an unknown runner might be told. The owner writes “Which Austin clinic helps runners return after an injury?” without adding the clinic’s name. The exact wording and location are part of the measurement.
Suppose the clinic uses the free test with its site, question and email. In the illustrative result, one assistant names the clinic positively, one links to its site only as a source, one names other clinics and one is unavailable. The right record is one positive name in three available responses, one source-only link and one unavailable response. The result is not a 25% forecast for every runner, and a missing answer is not an absent name.
- Confirm the question matches an actual service and market. Keep a copy of the exact sentence and do not add the business name just to obtain a favorable result. If you change the city or need, keep the revised question as a separate series.
- Enter the site, question and email into the free test. It requests the same question from four assistants and stores responses as they arrive. If a provider fails or a usage limit prevents a test, follow the message shown rather than counting an unavailable answer as negative evidence.
- Read the complete available answers, not just a count. Mark the clinic’s positive name in the prose, source-only links, other named businesses and any negative statements separately. A negative name is not a positive recommendation; a site link by itself does not satisfy the name criterion.
- Save the initial question and responses as the baseline. In the hypothetical result, inspect the cited pages and the clinic’s own service page. The source link may be a useful lead, but it does not reveal the entire process or prove why the assistant answered that way.
- Before making a change, ask a clinician to verify the service, area and booking instructions. Correct an inaccurate controlled page; if an independent directory is wrong, request a correction and retain its status. A request does not guarantee a listing or a future answer.
- If the owner later measures the same question again, save the new complete answers with dates and visible settings. Suppose two of four available answers now name the clinic positively. That is an observation in a later run, not proof that a specific change caused it or that the next customer will see the same answer.
If the clinic adds a second question about a different need after subscribing, its first measurement is that question’s own starting point; do not compare it directly to the first question’s baseline. If a provider is unavailable, retain the number of available answers beside the count. A linked site and a named business can be useful in different ways but should not be merged into one result.
Do not run the test with invented business facts, private patient details or a question for a service the clinic cannot provide. If the clinic has no capacity for additional appointments, solve that constraint first. The test is a measurement, not a repair and not a promise that an assistant will recommend the business.
What to do with the result
Read the available answers before deciding on a change. If the brand is absent, check the question, the accuracy of your page and the visible sources; absence alone does not explain the result. A name in the answer, a source-only link and an unavailable answer are distinct observations.
The free test saves the full answers to one question from ChatGPT, Gemini, Perplexity and Claude. After subscribing, the service works on the questions, requests inclusion where sources offer a route, and measures weekly. It costs $299 a month without lock-in, plus a one-time $599 bonus when a new positive name in a measured answer where the business was absent at baseline or an AI-referred visit recorded by connected Google Analytics occurs during the service. A team member checks the evidence; neither result is guaranteed.
Questions
Is the sayurname test really free?
Yes. The test is free and needs no payment card. Usage limits can apply; follow the message shown if a test cannot be started.
Which AI assistants does it check?
ChatGPT, Gemini, Perplexity and Claude, with the same settings for everyone.
Can I test more than one question?
The free test runs one question. After you subscribe you can add more.
If the AI does not name my brand, does the test fix that?
No. The test records a starting point. Correcting facts, requesting inclusion where possible and measuring again are separate work; none guarantees a future answer.