How does AI "recommend" brands?
Ask ChatGPT or Gemini a question such as "What project management tools do you recommend?" or "What skincare products help with irritated skin?" in Japanese, and the AI answers by naming specific brands.
So what criteria does AI use to choose those brands? Understanding how this works is essential for effective generative engine optimization (GEO).
Five factors in how LLMs choose brands
1. Frequency and context in the training data
Large language models (LLMs) learn from huge amounts of text. As a result, brands that are mentioned often in the training data tend to appear more often in AI answers.
However, raw frequency isn't everything; the context in which a brand is mentioned also matters. For example:
- Mentions in a positive context, such as "X is recommended" or "X is popular"
- Top placement in comparison and ranking articles
- Recommendations from experts and reviewers
- Mentions in industry reports and news articles
Conversely, if a brand is often mentioned in negative contexts (complaints, reports of problems, and so on), AI may also present the brand in a negative light.
2. Authority and credibility of sources
AI does not treat all sources equally. Information from highly authoritative sources tends to carry more weight.
- Coverage in major industry media: In Japan, outlets such as the Nikkei (a major Japanese business newspaper), ITmedia (tech news), and MarkeZine (marketing news)
- Reports from public bodies: Surveys by government agencies and industry associations
- Academic literature: Papers and research reports
- Trusted review sites: G2, Capterra, Kakaku.com (a major Japanese price comparison and review site), and others
3. Consistency of information (agreement across sources)
Because AI draws on multiple sources to generate its answers, it trusts information more when the same information can be confirmed across different sources.
For example, if only your own website claims you are "No. 1 in the industry" and no other source backs it up, AI is less likely to adopt that claim. Conversely, if multiple media outlets, review sites, and social networks give similar assessments, AI trusts that information and reflects it in its answers.
4. How relevant the brand is to the topic
AI gives priority to the brands most relevant to the topic of the question. This "relevance" is determined by the following factors:
- Category match: Whether the brand offers products or services in that category
- Depth of expertise: Whether the brand specializes in that topic or is a generalist
- Track record and case studies: Whether the brand has concrete results or case studies on that topic
What is interesting here is the Jobs-to-be-Done (JTBD) perspective. AI tends to recommend brands not by industry category but as solutions to the user's problem (their "job").
5. Recency of information
Especially with AI that searches the web in real time, such as Perplexity, the more recent the information, the more weight it tends to get. ChatGPT and Gemini also update their training data regularly, so the chances that your latest content will be reflected are increasing.
- Recent press releases and product updates
- Recent reviews and customer feedback
- The latest industry reports and trend articles
How the AI platforms differ
ChatGPT (OpenAI)
Its answers rely mainly on training data, and well-known brands tend to be mentioned more often. When it uses web search, the latest information is reflected too; without web search, it relies on what is in its training data.
Gemini (Google)
Because it is connected to the Google Search index, information from sites that rank high in Google search results tends to be reflected in its answers.
Perplexity
It searches the web in real time and generates answers that put weight on the latest information. Because it also shows the URLs of its sources, creating content that is likely to be cited is especially important.
Claude (Anthropic)
It is strong at analyzing long texts and tends to generate detailed, structured answers. Even niche brands may be mentioned if enough information about them is available.
Content strategy for getting recommended by AI
1. Establish your brand as an entity
Make sure AI can recognize your brand as an "entity" (a distinct, identifiable thing):
- Getting a Wikipedia entry and a Google knowledge panel
- Implementing structured data (Schema.org)
- Building out your company profile page
- Optimizing your Google Business Profile
2. Create "recommendation-friendly" content
Deliberately create content in formats that AI can easily draw on when it generates answers:
- Comparison articles: Make the differences from competitors clear with an "X vs. Y" format
- Top-picks lists: Aim to have your brand included in articles such as "5 picks recommended for X"
- FAQ format: Organize information around the questions users are likely to ask AI
- How-to articles: Explain concretely how to solve problems with your products
3. Earn third-party evaluations strategically
Beyond your own messaging, it is important to strategically increase evaluations from third parties:
- Encouraging user reviews on review sites
- Contributing articles to industry media and accepting interviews
- Publishing joint case studies with partner companies
- Entering industry awards
Monitoring mention share
Once you understand how AI recommendations work, the next step is to quantify how often your brand is actually recommended.
That is what the mention share metric is for. AIOPulse regularly sends Japanese-language prompts to multiple AI platforms (ChatGPT, Gemini, and Claude) and lets you track each brand's mention share over time.
Summary: becoming a brand that AI "chooses"
In a nutshell, AI recommends brands that are mentioned by trusted sources, frequently, in a positive context, and consistently.
This won't happen overnight, but by strategically building out your content and your brand's information ecosystem, you can steadily improve your standing.
To see how ChatGPT, Gemini and Claude answer Japanese users' questions about your brand, and where you can improve, contact us.
Related articles
- What is GEO? A guide from the basics (the big picture of AI search optimization)
- A guide to Knowledge Graph optimization for GEO (2026) (laying the groundwork for AI to recognize your brand accurately)
- A guide to building authority for GEO (2026) (becoming a brand AI chooses through E-E-A-T)
- What is mention share? How to measure and use it (the metric for comparing how often AI names each brand)