Be Recommended is an AI visibility tool that scores how ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews recommend your brand on a scale of 0 to 100, then tells you how to become the default recommendation. We built it at Inithouse after running 50+ real prompts across all five engines and watching the numbers come back.
The average brand we tested scored 31 out of 100. The top performers hit 80+. The gap between those two numbers is where we learned the most.
What the five engines actually do differently
Each AI engine retrieves, ranks and presents brand recommendations in its own way. Here is what we found after running structured prompts across all five:
| Engine | Retrieval method | How it recommends | What surprised us |
|---|---|---|---|
| ChatGPT | Web search + training data | Ranks by perceived authority, cites sources inline | Picks up Dev.to and Indie Hackers posts within days |
| Claude | Web fetch + search index | Cautious, flags missing third-party validation | Fetches product pages directly, penalizes thin content |
| Perplexity | Real-time search, multiple sources | Shows source cards, mixes owned and third-party | Volatile between runs, search depth varies |
| Gemini | Google index + grounding | Builds from indexed pages, adds its own framing | Embellishes product descriptions with expected category traits |
| Google AI Overviews | Search results + knowledge graph | Summarizes top-ranked pages | Cannibalizes clicks from traditional search results |
That table came out of months of running the same prompts repeatedly. Not a one-off test.
The 31-point average tells a specific story
Most brands we tested had basic web presence: a homepage, maybe a blog, social profiles. They scored between 20 and 40. The AI engines could find them but had little to say about them.
The brands scoring 80+ shared a few traits. They had content published on multiple platforms. They had third-party mentions (reviews, comparisons, listicle features). And the information about them was consistent across sources.
The score gap maps to a specific problem: when an AI engine can only find your own website talking about your product, it treats you as an unverified claim. When independent sources corroborate what you say, your score climbs.
We see this across our own portfolio. At Inithouse we run about 17 products, from Watching Agents (an AI prediction and monitoring platform) to Audit Vibe Coding (a code audit tool for AI-generated projects). Products with published third-party content consistently outscore products that rely only on their own site.
How Be Recommended compares to existing tools
Tools like Otterly.ai track AI mentions and monitor how often brands appear in AI-generated answers. They focus on tracking over time.
Be Recommended takes a different cut. Instead of ongoing monitoring dashboards, it runs a structured audit across five engines using 50+ prompts tailored to your category, scores the result on a 0-100 scale, and gives you a prioritized action plan. The output is a report, not a dashboard.
Both approaches have their place. Ongoing monitoring makes sense once you know your baseline. The initial audit makes sense when you need to figure out where you stand and what to fix first.
Three patterns we did not expect
AI engines disagree with each other regularly. A brand can score 70 on ChatGPT and 20 on Claude in the same week. The engines use different retrieval pipelines, different source weighting and different trust signals. A single-engine check tells you almost nothing about your overall AI visibility.
Published content migrates between engines unpredictably. A blog post published on Dev.to might show up in Perplexity citations one week and disappear the next, then appear in Claude results a month later. The engines do not maintain stable indexes the way Google Search does.
Your own content can work against you. If your published materials contain inconsistent claims (different feature lists on different platforms, outdated pricing, contradictory positioning), the AI engines pick up the contradictions and lower confidence in your brand. Consistency across sources matters more than volume.
What we actually measure and how
Be Recommended generates a composite score from prompt-level signals across all five engines. For each prompt, we check whether your brand appears in the response, what position it holds relative to competitors, whether the AI cites your owned sources or third-party sources, and whether the description is accurate.
The 0-100 score is not a vanity metric. Brands that moved from 30 to 60 in our tests did so by publishing consistent information on high-authority platforms, getting featured in comparison articles, and fixing contradictions in their existing content.
If you want to check where your brand stands across ChatGPT, Claude, Perplexity, Gemini and Google AI Overviews, you can run an audit at berecommended.com.
We build and run AI products at Inithouse. Be Recommended grew out of the AI visibility tracking we do for our own portfolio.
Top comments (0)