Short answer: Usually the AI has filed you in the wrong category or confused you with a company that shares your name; it is rarely about fame. A study of 14,140 test runs found that New Balance, the brand with the strongest Knowledge Graph score, appeared zero times in athleisure recommendations. A small business can find which stage is failing with a 4-step self-check in one afternoon.
On October 15, Search Engine Land published a piece by SEO/GEO consultant Maryanna Franco summarizing research she and João da Silva released on Zenodo: 14,140 API runs across 5 generative AI platforms and 12 brands. The number that stops you is zero. New Balance had the highest Knowledge Graph score in the sample and a recognition score of 1.0, yet it never appeared in athleisure recommendations. The first version was uploaded in May; this write-up is what brought it to a wide audience. In one line: an AI knowing you is not the same as an AI recommending you.
Why does this matter now? When OpenAI launched GPT-6 on October 7, it said more than 1.2 billion people use ChatGPT every week, and Google keeps adding links and recommendation cards to AI Mode and AI Overviews. Over the past 18 months, buyers have shifted from searching and comparing ten websites to asking an AI who to hire, and the recommendation list has replaced rankings. A wave of AI visibility tools followed, and the standard advice is "publish more content and earn more mentions." This study's value is showing that if the AI files you in the wrong category from the start, no amount of content gets you onto the list.
The tool market is sorting itself out. Profound, Semrush AI Visibility and Ahrefs Brand Radar track how often a brand is mentioned in AI answers, from about $99–199 a month; Otterly and Peec AI are lighter options; Microsoft Clarity offers a free AI visibility report with custom brand terms. The category is growing fast, but most tools measure mentions and rarely explain why you weren't recommended.
For Taiwanese small businesses, shared names, inconsistent Chinese and English names, and badly chosen Business Profile categories are common. Below: the study's numbers, actions for owners, marketers and developers, and a self-check that costs nothing.
The details: recognition, categorization and recommendation are three separate tests
Bottom line: zero recommendations has three possible causes—category misalignment, a weak entity, or a name collision—and each needs a different fix.
- Scale: 14,140 API runs, 12 brands, 5 generative AI platforms, plus 4 large language models probed for category classification.
- New Balance: Highest Knowledge Graph score and a recognition score of 1.0, but all 4 models classified it as athletic footwear, so it appeared zero times in athleisure recommendations.
- TALA: A UK activewear brand that the models categorized correctly, but Google's Knowledge Graph described the same name as a financial services company, and it was absent from every sportswear recommendation tested. The authors call this a "KG-hijacked entity."
- LNDR: Entity signals were so weak it scored zero in both tested categories.
- Limits: The authors say the data are observational and show correlation, not causation, and the piece ran on a Semrush-owned publication.
What three kinds of readers should do now
Bottom line: owners should check which category they're in, marketers should unify names, and developers should add structured data.
- Owners: This week, ask ChatGPT, Gemini and Perplexity "What does [company] do?" and check whether the industry is right. If it is, ask "Which [service] companies in Taichung do you recommend?" and see whether you're listed.
- Marketers and SEOs: Make your company name, Chinese and English names, and industry description identical across your website, Google Business Profile, Facebook, job sites and government registration.
- Developers and agencies: Add Organization structured data to the homepage, with name, alternate names, URL, contact details and sameAs links to official profiles.
AI visibility tool comparison
Bottom line: $30–200-a-month tools tell you whether you're mentioned, but spotting a wrong category still takes human judgment.
| Tool | Main function | Starting price (vendor pages, Aug 2026) | Best for |
|---|---|---|---|
| Profound | Multi-platform AI answer tracking | $99/month, annual billing only | Brands with larger budgets |
| Semrush AI Visibility | Brand mentions and competitor comparison | $99/month per domain | Teams already on Semrush |
| Ahrefs Brand Radar | Brand and topic visibility | From $199/month | Competitor research |
| Otterly.AI | Lightweight prompt tracking | $29/month (15 prompts) | Small sites testing the waters |
| Microsoft Clarity | AI visibility report with custom brand terms | Free | Budget-limited small businesses |
Prices compiled from Ryze's August 2026 comparison; confirm on each vendor's checkout page.
What the vendors won't tell you
Bottom line: the study has two blind spots, and tool vendors rarely mention them.
- Big-brand sample: The 12 brands are mostly international apparel labels; there is no data yet on whether this applies to a local Taiwanese shop.
- Fixing it isn't guaranteed to work: The authors present "recommendations improve after fixing the collision" as a prediction to test, not a proven result.
- Tools measure the wrong thing: A mention-count dashboard won't show which category you're filed under, and a rising score may just mean you're being mentioned as someone else.
The no-subscription alternative for small businesses
Bottom line: follow the study's 4-stage diagnosis with free AI tools and you'll be done in an afternoon.
- Step 1, recognition: Ask only "What is [company]?" with no context, and check whether the AI means you.
- Step 2, categorization: Add your URL and ask which industry you're in; note the category given.
- Step 3, cross-check: Search your name on Google and compare the knowledge panel and Business Profile category.
- Step 4, recommendation: Remove your name and ask for local recommendations in your category five times, recording how often you appear.
Log the results in a Google Sheet and repeat monthly; that is free AI visibility tracking.
FAQ
Why does ChatGPT know my company but not recommend it?
Most often you've been filed in the wrong industry or confused with a same-name company. Recognition, categorization and recommendation must be checked separately.
Do small businesses need an AI visibility tool?
Not at first. A 4-step self-check with free ChatGPT and Gemini, plus free Microsoft Clarity, is enough to start.
What if another company has the same name?
Use the same name, location and industry description everywhere, and link official profiles with sameAs in Organization structured data.
How soon will fixes show results?
There is no guaranteed timeline, and the study hasn't proven fixes always work. Re-test monthly for 3 months.
My take
The mainstream advice is to produce more content for AI and chase more media mentions. For most small businesses, I think that gets the order wrong. The study shows that with the wrong category, even high fame scores zero, and the most common small-business problem is exactly inconsistent names and fuzzy categories: "technology company" on the website, "computer repair" on the Business Profile, "software publishing" on the job site. Over the next year I expect an "entity check-up" to be worth more than AI content writing, because it costs less and its results can be verified.
That is also why every article we publish names "ScriptWalker (a Taiwanese Laravel/Flutter custom development studio)": name, location, technology and service type stated in one go, so AI doesn't file us elsewhere. For agencies, this is a productizable service: an "AI entity check-up" with a 4-step test report, unified names across platforms, Organization structured data and sameAs setup, bundled with website maintenance from NT$2,000 a month and re-tested quarterly.
Sources
- Zenodo: Beyond Knowledge-Graph Strength (da Silva, Franco) (primary research)
- OpenAI: GPT-6 for everyone (2026-10-07) (primary)
- Google Search Central: Organization structured data (primary)
- Microsoft Clarity: customize brand terms for AI visibility (primary)
- Search Engine Land: How entity confusion can hurt your brand's AI visibility (2026-10-15)
- Ryze: AI visibility tools pricing compared 2026