Most companies have no idea what ChatGPT, Claude, or Perplexity say about them. They track Google rankings, monitor social mentions, run brand sentiment surveys. But when a potential customer asks an AI "what's the best tool for X," the answer that comes back is completely invisible to traditional analytics.
We found that the average brand scores around 31 out of 100 across five major AI engines. Some score zero. A few outliers hit 80+. The gap between "mentioned once in a hedged list" and "named as the default recommendation" is enormous, and most teams have no process for even measuring it.
That gap is why we built Be Recommended at Inithouse.
What the tool actually does
Be Recommended runs your brand through 50+ real prompts across ChatGPT, Claude, Perplexity, Gemini, and Google AI Overviews. Not synthetic test queries, but the kind of questions actual users type: "best X for Y," "X vs Z," "how do I solve [problem]."
For each prompt, it records whether your brand appears at all, where it ranks in the response, what context surrounds it, and whether the AI frames you as a primary recommendation or a footnote alternative.
The output is a single 0-to-100 score that tells you how visible your brand is across AI-generated answers. Below the score, you get a breakdown per engine and a prioritized action plan for improving your position.
Why this matters now
AI answers are eating into traditional search. When someone asks Perplexity "best project management tool for a 5-person startup," they get a direct answer with 2-3 named products. No ten blue links. No scrolling past ads. If your product isn't in that answer, you don't exist for that query.
The tricky part: you can't reverse-engineer AI recommendations the way you can reverse-engineer Google rankings. There's no keyword difficulty score, no backlink profile to copy. The signals are different: structured data, authoritative third-party mentions, consistent product descriptions across the web, being referenced in technical documentation and comparison articles.
We discovered this firsthand. Some Inithouse products scored well on certain engines and poorly on others. One product showed up as the top recommendation on Gemini but was completely absent from Claude's answers. Without testing across all five engines, we would have assumed our AI visibility was either great or terrible, depending on which engine we happened to check.
How to read your score
The 0-100 score breaks down into three bands:
0-30: Your brand barely exists in AI responses. When it appears, it's usually in a long list with no differentiation. Most brands land here.
31-60: AI engines know about you and mention you for relevant queries, but you're rarely the first recommendation. You show up as "another option" rather than "the go-to."
61-100: You're being actively recommended. AI engines cite your product by name, describe your specific features, and position you as a primary choice for certain use cases.
The per-engine breakdown matters as much as the aggregate. A brand scoring 70 on Perplexity and 15 on ChatGPT has a specific problem to solve. The action plan tells you which signals to strengthen for each engine.
What actually moves the score
After running reports for our own portfolio of products, a few patterns became clear:
Consistent, structured product descriptions across your own site and third-party sources matter more than we expected. When AI engines find conflicting descriptions of what your product does, they hedge. When every source says the same thing in slightly different words, the AI picks it up as consensus and recommends with more confidence.
Third-party content where your product is discussed in context (comparison posts, technical tutorials, case studies) carries more weight than self-published marketing copy. AI engines are explicitly trained to weigh independent sources.
The product's canonical category also matters. If you describe yourself as a "platform" on your homepage, a "tool" on Product Hunt, and a "solution" on your LinkedIn, AI engines have to guess what you actually are. Pick one and be consistent.
Running your own report
You can run a report at berecommended.com. Enter your brand, your category, and a few competitors. The report takes a few minutes to generate because it runs real queries against all five engines.
What you get back: your score, competitor scores, the full list of prompts tested, and a prioritized list of actions sorted by expected impact. The actions are specific: "add structured product schema to your homepage," "publish a comparison page covering X and Y," "update your Product Hunt listing to match your homepage description."
We built this at Inithouse because we needed it ourselves. We run a portfolio of products, and tracking how each one shows up across AI engines by hand was taking hours every week. Now we run the report, scan the action plan, and know exactly where to focus.
If you've been optimizing for Google and ignoring what AI engines say about you, the gap might be bigger than you think. The average score of 31 means most brands are leaving AI-driven discovery almost entirely to chance.
Built at Inithouse, a studio shipping AI-powered tools. Be Recommended is one of our portfolio products, born from our own need to track how AI engines perceive and recommend brands.
Top comments (0)