AI search visibility is how often and how prominently AI engines name your brand when people ask about your category. MentionRadar runs your prompts weekly in ChatGPT, Gemini, Perplexity, Claude and Google AI Overviews and scores each answer for mention, position, recommendation and cited pages.
“What is the best project management software for a small agency?”
For small agencies, three options come up most often. 1. Brightpath, flexible boards and simple pricing. 2. Fieldnote, built around client projects, time tracking and approvals. 3. Lumen CRM, a good fit if sales and delivery live in one place.
AI visibility score
64
out of 100
An AI answer is short. Being in it, high in it and endorsed by it are different things, and each one is recorded separately for every prompt and engine.
The engines do not work the same way, which is why one prompt can name you in Perplexity and ignore you in Claude. A general overview, as each engine behaves by default for most users.
| Engine | Live web search | Shows sources | What it means for your visibility |
|---|---|---|---|
| Perplexity | Searches the web for nearly every answer | Numbered citations on each answer | Reacts fastest to new reviews, articles and threads. Good place to see the effect of fresh content. |
| ChatGPT | Uses web search when the question calls for it or the user turns it on | Links sources when it searches | A mix of model knowledge and recent pages. Visibility tends to move gradually. |
| Gemini | Can ground answers in Google Search results | Shows sources for grounded answers | Pages that are strong in Google results have an advantage. |
| Claude | Answers from model knowledge unless web search is enabled | Cites sources when it searches | Changes slowly. Broad, consistent presence on reputable pages matters most. |
| Google AI Overviews | Generated on the results page from pages in the Google index | Links to the pages it draws from | Closely tied to classic organic search, so existing SEO work carries over. |
Engines change their behaviour over time, and settings differ between free and paid accounts. MentionRadar records the answer each engine returns to your exact prompt, so you track the result rather than a theory of how it was produced. Google AI Overviews get a dedicated view on the page about AI Overview tracking.
Overall score
64
+5 vs last month
Every answer earns points. A recommendation is worth 100. A mention in a list without a recommendation is worth 100 minus 15 for every place below the first, with a floor of 40, so a fourth place earns 55 and anything from fifth place down earns 40. An answer that does not name you earns 0.
The score is the average of the latest run of every prompt and engine pair from the last 8 days. Take four pairs: Perplexity recommends you (100), ChatGPT lists you fourth (55), Gemini lists you seventh (40) and Claude does not name you (0). The score is 195 divided by 4, which is 48.8.
Because only the latest run of each pair counts, the number reflects what engines say now, not an average of the last quarter. The per-engine breakdown shows where the gap is. In the screen on the left, Claude pulls the overall score down, which usually points to thin presence on the reputable pages a model learns from.
Position over time has its own chart in the AI rank tracker, with your brand plotted against competitors week by week.
Run through this list once with the first weekly results in front of you. It turns a score into a short list of pages to work on.
Improving the result is the work of answer engine and generative engine optimization. Measuring it is what MentionRadar does. For the optimization side, read about our AEO tool and the generative engine optimization tool view.
AI search visibility describes whether AI engines name your brand when someone asks a question about your category, where in the answer you appear, whether you are recommended and which pages the engine relies on. It is the AI counterpart of organic search rankings, measured across answers instead of result pages.
You choose prompts that match real buying questions. Each prompt runs weekly in every engine on your plan, and the answer is parsed for your brand, its position in any numbered list, a recommendation and competitor names. The AI visibility score averages the latest run of every prompt and engine pair from the last 8 days.
Starter tracks ChatGPT, Perplexity and Google AI Overviews with 15 prompts a week. Growth, Agency and Enterprise add Gemini and Claude, with 40, 150 and 300 prompts a week. There are no per-engine add-ons.
Engines differ in whether they search the live web for a question, which index they search, how they choose sources and how much they lean on what the model learned in training. An engine that searches the web reflects recent pages quickly, while an answer from model knowledge changes slowly. That is why visibility has to be tracked engine by engine.
Every tracked prompt runs once a week in each engine on your plan. AI answers vary from run to run, so a weekly cadence over several weeks gives a more reliable picture than a single daily snapshot. Web mentions, by contrast, are collected every 15 minutes.
Usually yes, through the same pages engines cite: comparison articles, review sites, documentation and discussions that name your brand clearly. MentionRadar shows which domains the engines cite for each prompt on Growth and higher plans, so you know which kind of page to work on.
Run a scan and see a real AI answer for your category next to your latest web mentions, in about a minute and before you create an account.