How do I check if my website is visible in AI search results?
How we check whether a client's website is visible in AI search: manual tests, Search Console and logs, then a tracker that repeats the test and shows why.
Every new client gets the same first question from us: is the site visible in AI search today? We answer it in layers, because one layer alone misleads. Here is the sequence we use, which you can follow yourself.
Layer one: ask the engines
We pick five to ten questions a real customer would ask, not brand-name queries. We run each one in ChatGPT, Google (AI Overviews, and AI Mode where available), Perplexity, Gemini and Copilot, in fresh sessions. For each we record whether the brand is named, whether the site is cited, and which competitors are named.
This takes an hour and produces a rough picture. It also produces our first disagreement with the client's expectations, because the answers differ from run to run. That is normal. It is also why we do not treat this layer as measurement.
Layer two: check what the site already tells you
Search Console. As Google describes it, appearances in AI features are counted within the overall Web search totals of the Performance report. We look for question-style queries with impressions and for pages that earn clicks on informational topics. We check the documentation again at the start of each engagement, since it changes.
Bing Webmaster Tools. Copilot draws on Bing, so indexing problems there are an early explanation for gaps.
Analytics. We check referral sessions from AI assistants.
Server or CDN logs. We look for AI crawlers: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Claude-User, Claude-SearchBot, PerplexityBot, Perplexity-User and Google-Agent. If a key page is never requested, we know where to start.
In our experience, this layer produces the most useful early finding: a blocked crawler or an error on a page the client assumed was fine.
Layer three: free tools
The free tools from Promptwatch are good for a first pass. The AI Visibility Report is a no-login scan. The Agentic Readiness Check looks at how ready a site is for AI agents. The GEO Inspector Chrome extension shows ChatGPT fan-outs and which sources were fetched versus cited, running locally. The llms.txt and robots.txt generators are handy if the audit finds gaps.
Layer four: a tracker that repeats the test
Manual checks cannot be repeated daily across countries and models. For that we use Promptwatch, because it does the repeating and then helps explain the result.
- Monitors run daily per model, country and language, and report Visibility Score, Share of Voice, Sentiment and Position in Answer.
- Citation analytics shows which pages and domains are cited, with Reddit, YouTube and LinkedIn reports and offsite mentions.
- Agent Analytics gives real-time, IP-verified crawler logs and a citation rate per page. A page that is crawled and never cited needs a content fix. A page that is never crawled needs an access fix.
- Search Insights overlays Google Search Console and Bing Webmaster data, including Microsoft Copilot traffic, on crawls and citations, which puts the free data in the same view.
- Visitor analytics attributes AI-referred traffic and conversions, using a lightweight script, the GTM template or the WordPress plugin.
What we do with the findings
We sort every page into one of four groups.
- Not crawled. Check robots.txt, firewall rules, rendering and sitemap. Fix before anything else.
- Crawled and not cited. The page is reachable but does not answer the question well, or a competitor's does it better. Content gap analysis with coverage scores helps here.
- Cited and not converting. The traffic comes and does nothing. This is a landing-page problem, not a visibility one.
- Cited and converting. Protect it. Refresh it before it decays.
The Action Board gives a prioritized to-do list, and a Discovery agent investigates the data weekly and suggests up to five evidence-backed actions per run.
Choosing a plan
The free Explore plan covers 10 prompts on ChatGPT only, which is enough to see how the report works. Essential at $95 a month includes all active models with 50 prompts. Crawler logs are not in Essential. They begin on Professional at $245 a month, which is the tier we recommend as soon as the audit shows crawler questions. For agency work, the Kick-off, Growth and Scale plans at $199, $399 and $799 include unlimited projects and prompts.
What a visibility check will not do
It will not promise a recommendation, and it will not show a stable ranking because AI answers do not have one. What it gives you is a baseline and a way to tell whether the next change helped. Run the checks for two weeks before you change anything, then compare.