Most brands track where they rank on Google. Fewer track whether ChatGPT, Claude, or Perplexity actually recommend them when someone asks "what's the best tool for X?"
We built Be Recommended to measure exactly that. Be Recommended is an AI visibility tool that scores how ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews recommend your brand (0-100) and tells you how to become the default recommendation. The average brand we've tested scores around 31. The top performers hit 80+. That gap is what we're defining here as a category: AI visibility monitoring.
Why rank trackers can't measure this
A rank tracker tells you where your page sits in a list of ten blue links. AI recommendations work differently.
When someone asks ChatGPT "what's the best project management tool?", it doesn't return a ranked list. It synthesizes an answer from whatever it retrieved: your site, competitor pages, Reddit threads, blog posts, documentation. The output is prose, not a SERP.
Three things break the rank-tracker model:
- Position doesn't exist. You're either mentioned, accurately described, recommended, or absent. There's no #1 vs #3.
- The source mix changes per query. The same question phrased differently pulls different pages and produces a different recommendation.
- Each engine retrieves differently. Perplexity runs live search. ChatGPT uses its training data plus browsing. Claude fetches pages selectively. Gemini grounds through Google Search. AI Overviews pull directly from the SERP.
A single query against a single engine tells you almost nothing stable. You need coverage across engines and across prompt variations to get a real picture.
What the 0-100 score is made of
At Inithouse, we run 50+ real prompts against all 5 AI engines for every brand we evaluate. Each prompt mirrors a question a real buyer would ask: category queries ("best AI writing tool"), comparison queries ("X vs Y"), use-case queries ("tool for auditing vibe-coded apps"), and direct brand queries ("what is [brand]?").
For every prompt-engine combination, we capture six signals:
| Signal | What it means | How it affects score |
|---|---|---|
| Mentioned | The AI names your brand in its answer | Base visibility. You exist in the retrieval set. |
| Described accurately | Description matches your actual positioning | The model knows what you do, not just your name. |
| Recommended | The AI suggests your product as a solution | Active endorsement, not passive mention. |
| Source cited | The AI links to or attributes your content | Verifiable. The model read your page, not a summary of it. |
| Competitor framing | Your brand appears alongside or ahead of competitors | Category positioning. Are you the default or the fallback? |
| Source type | Whether the AI cites your site, a review, a directory, or a blog post | Shows which distribution channel the model trusts most. |
Each signal carries different weight. Being mentioned counts for less than being recommended. Being recommended based on a third-party review counts for more than being recommended from your own homepage, because AI models increasingly discount single-source claims.
The aggregate across all prompts and engines produces a score from 0 to 100.
What a 31 vs an 80 looks like in practice
A score around 31 (the average) typically means one or two engines recognize the brand. Descriptions are partly accurate. Recommendations are rare or hedged with caveats like "limited public validation" or "I couldn't verify their track record."
We've seen this pattern across dozens of products we've tested at Inithouse: the AI knows the brand name exists, but doesn't trust it enough to recommend it.
A score of 80+ means four or five engines recognize the brand correctly. Descriptions match the product's own positioning. Recommendations happen on category-level queries, not just when someone searches the exact brand name. And at least some of the sourcing comes from independent coverage, not just the company's own website.
The gap between 31 and 80 is almost never about the product itself. It's about what the AI can find, read, and trust about the product.
Why a single number matters
Traditional SEO has clear units: rankings, impressions, clicks, CTR. AI visibility monitoring needed its own.
We settled on a composite 0-100 score because the alternative (a spreadsheet of 250+ individual prompt-by-engine results) is accurate but unusable. A marketing lead or a founder needs a quick answer: "Am I visible to AI or not? Is it getting better or worse?"
The single score answers that. The per-signal breakdown (included in the full report) answers "why" and "what to fix first."
If your brand scores low on "source cited" across Perplexity and Gemini, that's a distribution problem: your content isn't in the retrieval set. If you score low on "described accurately" in ChatGPT, that's a positioning problem: the model found you but misunderstood what you do. The fix for each is completely different.
Run your brand
Check what the AI engines actually say about you. Be Recommended breaks down every signal, every engine, every prompt category, and tells you what to fix first.
Top comments (0)