Check your visibility by sampling, not by chasing a rank. Search a handful of real customer questions in Google and Maps, ask the same questions of two or three AI assistants, and write down what appears, which pages are cited and the date. Repeat monthly. Answers vary between people, places and days, so treat each result as one sample.
Three places to check
- Google search and Maps. Search the phrases a customer would use, with and without your town. Note whether your site, your Business Profile and review listings appear, and who else does.
- Your listings. Business name, phone number, hours, service area and categories should match everywhere they appear: your website, Google Business Profile, Apple, Bing, and directories.
- AI assistants. Ask a few of those same customer questions in assistants such as ChatGPT, Gemini, Perplexity and Claude, with web search on and off, and record what they say.
What to record each time
| Field | Why it matters |
|---|---|
| Date and who ran it | Results change over time; the log is only useful if dated. |
| Platform and mode | For example Google, Maps, ChatGPT with web search on. Behavior differs between modes. |
| Exact prompt or query | Small wording changes alter results. Reuse the same wording each month. |
| Location and device | Local results depend on where the search is made. |
| Businesses named | Shows who you are being compared with. |
| Sources cited or linked | Shows which pages the assistant relied on, so you know what to improve. |
| Whether your business appeared and how it was described | Check the description for errors. Wrong facts are a fix list. |
Start with five to ten prompts. Typical shapes: "best [service] in [town]", "who can help with [problem] near [town]", "how much does [service] cost", "is [your business] any good". Avoid prompts that name you in a flattering way, because they test the assistant's politeness, not your visibility.
How the platforms find your pages
Each search engine and assistant finds pages through its own crawlers, and each vendor publishes its own documentation. Before you change anything, check your robots.txt file and read the crawler documentation of every platform you care about. Blocking a search crawler can keep your pages out of that platform's search results, and permission to use your pages for model training is usually a separate setting from permission to be found in search.
| Check | How | Why it matters |
|---|---|---|
| robots.txt rules | Open yourdomain.com/robots.txt and read every User-agent group, including any blanket Disallow. | A rule meant for one crawler can block others. Rules are easy to leave behind after a site migration. |
| Page-level blocks | View the page source or use your Search Console inspection tool and look for a noindex robots meta tag or header. | A noindex tag keeps a page out of search even when crawling is allowed. |
| Page status and content | Confirm important pages return a normal 200 status, show their text without a login, and have a canonical URL that points to themselves. | A page that errors, redirects oddly or loads its text only through scripts may not be read as intended. |
| Vendor documentation | Read each platform's current crawler page and note which crawlers it lists and what each is for. | Names and purposes change. Rely on the vendor's current page, not on a blog post, including this one. |
For example, Anthropic's Help Center article on its crawlers (dated April 7, 2026) describes three robots, ClaudeBot for model training, Claude-User for fetching pages when a person asks Claude a question, and Claude-SearchBot for improving search result quality. It says its robots follow robots.txt directives, and that disabling Claude-User or Claude-SearchBot may reduce your site's visibility in user-directed web search or search results. Treat that as one platform's description, not a promise about any outcome.
What to fix first
- Facts that disagree. Make name, phone, address or service area, hours and services identical across your site and listings.
- Pages that answer the question. Give each service its own page that states what you do, who it is for, where you work, what is included and how to start, in plain text a crawler can read.
- Real reviews and mentions. Ask satisfied customers to review you honestly. Do not buy reviews.
- Crawlability. Make sure important pages are not blocked, not hidden behind scripts, and appear in your sitemap.
- Errors in AI descriptions. If an assistant gets a fact wrong, find the source it cited and correct it there.
What cannot be promised
- No one can guarantee a ranking or that an AI assistant will mention your business.
- Files such as llms.txt and markup such as FAQ schema are optional extras. We do not treat either as a way to secure a mention, and no one should sell them as one.
- Results vary by person, location, wording and time, so one check proves little. A pattern across several dated checks is more useful.
Measuring over time
Search Console and Bing Webmaster Tools show impressions and clicks for your own site. Bing Webmaster Tools may offer additional reporting; check what is available in your own account. In analytics, some assistants pass a referral source, but many answers send no visit at all, so treat referrals as a sample. Separate branded searches (people already looking for you) from non-branded ones, because they tell you different things.
If you want help
QS Digital's SEO and AI search visibility checks cover this method for a business, with findings reported as dated observations and a prioritized fix list. We are based in Vineland, NJ and work remotely. We have not yet published our own test log, and we will not present these checks as results until we have run and dated them.
Sources and evidence
- Anthropic Help Center, Does Anthropic crawl data from the web, and how can site owners block the crawler?. Accessed 2026-10-10. Page dated April 7, 2026.