AI search visibility: find out what the assistants say about you
Buyers now ask ChatGPT, Perplexity, Google AI Overviews and Claude before they search. We build the monitoring that shows whether you are mentioned, cited and recommended, and what to change when you are not.
When an assistant answers "which platforms offer badge printing and lead retrieval?", the buyer may never visit anyone's website. Three names are listed, one is chosen for a demo, and the other vendors never learn they were in the running. Nothing appears in Search Console. Nothing appears in your CRM.
Referrals from chatgpt.com or perplexity.ai do show up in analytics, but they are the minority of cases: the ones where the buyer clicked a citation. Being named without a click is still influence, and it is invisible unless you measure it directly.
Measuring it directly means asking the engines the questions your buyers ask, on a schedule, and recording what comes back. That is what we build.
5Engines monitored by default
50–200Prompts per set
×3Runs per prompt to average variance
MonthlyTrend and gap report
How it works
The monitoring loop
The loop runs monthly. Step 3 uses a model with rules to parse answers; step 5 feeds work back into content and markup. Raw answers are stored so you can audit any score.
What gets cited
Content formats that engines lift most often
Across the audits we have run, the pattern is consistent: engines cite passages that answer a question completely in one place. These formats do that.
Format
Why it gets cited
Example on this site
Comparison tables
Rows map directly onto "X vs Y" questions; the model can lift a cell without paraphrasing
Your mention and citation rates per engine, the competitors who appear instead, and the URLs engines currently use for you. Two weeks.
Monitor
A scheduled job that re-runs the prompt set, stores raw answers and scores them. Hosted by us or in your cloud. Built on the same tooling as our LLM applications.
Monthly gap list
Prompts lost, prompts gained, pages cited instead of yours and specific fixes. Connected to your analytics so visibility sits next to leads.
When the engine gets you wrong
Monitoring also catches hallucinated claims: a wrong founding year, a product you discontinued, a price from 2019. These are usually traceable to a stale page or an inconsistent third-party listing, and they are fixable.
Process
From first run to monthly cadence
Prompt workshop
Half a day with your sales and support teams to collect the questions buyers really ask, grouped by funnel stage.
First run
Three runs per prompt per engine. Raw answers stored. Parsed with a model plus rules, spot-checked by a person.
Baseline review
A one-hour walkthrough: where you stand, who wins today, the ten fixes with the best ratio of effort to visibility.
Schedule
The monitor runs monthly, or weekly during a campaign. Alerts when a prompt flips from cited to absent.
Why can I not see AI visibility in Google Search Console?
Search Console reports Google search impressions and clicks, including some AI Overview traffic, but it cannot tell you what ChatGPT or Perplexity said about you. Citations that do not lead to a click leave no trace in your analytics. You have to ask the engines directly, repeatedly, and record the answers.
AI answers change every time. How can you measure them?
By sampling. We run each prompt several times per engine, on a schedule, and report rates rather than single results: mentioned in 7 of 10 runs, cited in 4 of 10. Trends over weeks are reliable even when individual answers vary.
Which engines do you monitor?
ChatGPT with search, Perplexity, Google AI Overviews and AI Mode, Microsoft Copilot and Claude-style assistants with web access. The set is adjusted to where your buyers actually are, which we check with a short survey or your sales team.
Do you use a third-party tool or build your own?
Either. Several commercial tools exist and we will recommend one if it fits. For clients who want raw answers stored, custom scoring or integration with their data warehouse, we build a monitor using the engines' APIs and browser automation. The code is yours.
What do we do with the results?
Each month you get a list of prompts where you lost or never had presence, the pages that were cited instead, and specific content, markup or entity fixes. The fixes feed our GEO programme or your own team.
Ready to make AI work in production?
Tell us about your systems, your data and your constraints. We reply within one business day with a straight answer and a suggested next step.