What does AI trust?
Generative AI learns from vast amounts of information on the Internet, reads the context, and generates natural sentences. However, the fact that AI not only collects information, but also has a decision axis of “which information to trust,” has extremely important implications for companies designing their Web strategies.
For example, ChatGPT and Google-generated AI tend to preferentially cite “primary” or “official” information available on the Web, or highly reliable media (government agencies, educational institutions, major media). Furthermore, as with search engines, domain authority (high reliability) and the presence or absence of structured data are also used to make decisions.
In other words, reliable information for an AI is one that has the following elements
- Must be primary information (original content)
- The originator must be clear (e.g., company, researcher, expert, etc.)
- Clear structure and context, and clarity of intent
- Frequently updated and up-to-date
- Must have references or links from external sites
Must have references or links from external sites
The role of official company information
From the perspective of a BtoB company, there are an increasing number of situations where the generating AI takes up the company name, service details, and representative information. For example, when we ask ChatGPT, “What is 00 Corporation?” where does the AI pull the information from?
An important case here is when a company’s official website, press release, or recruitment page are used as “information sources”. In particular, the following information can be useful data points for AI
- Company profile (location, year established, nature of business, name of representative)
- Service introduction page (features, prices, introduction results, etc.)
- Press releases (new services, partnerships, awards, etc.)
- Recruitment information (vision, culture, evaluation system)
- Messages and interviews with representatives
Whether or not such information is organized as primary information about the company will determine whether or not AI will trust it and use it in its answers. In particular, being structured (markup such as schema.org) and organized in an easy-to-understand manner will facilitate understanding by the AI.
Significance of original knowledge dissemination
To become a “frequently cited” source of information by a generating AI, you must be unique and differentiated from other companies. In other words, it is extremely important to disseminate primary knowledge that is unique to your company.
Primary knowledge in this context is the following
- Detailed reports on your company’s successes and achievements
- Share specialized processes and know-how (e.g., UI/UX design approaches)
- Interview and dialogue content (own members, clients, partners)
- Independent research, survey results and market analysis
- Knowledge and tips gained in practice (AI utilization, marketing improvement, etc.)
Continuous dissemination of these in the form of owned media, blogs, knowledge pages, and case studies will increase the company’s credibility and result in more frequent references from the AI.
In addition, the following innovations can be added to the format to increase its adaptability to the generated AI:
- Organize in FAQ format (i.e., GEO compliant)
- How to~” type know-how articles
- Clear organization with charts and lists
- Article design that includes evidence (e.g., research findings, clearly cited sources)
In other words, not only the “quantity” of primary information, but also its “structure,” “quality,” and “continuity” determine its reliability for AI.
GEO value of owned media
Another important perspective here is Generative Engine Optimization (GEO). Whereas traditional SEO aims for high search engine rankings, GEO is a design philosophy that aims to get cited by generative AI.
In GEO, the following designs are recommended
- Add structured data such as schema.org to pages
- Described as “meaningful cohesion” such as FAQs, how-to, etc.
- Independent pages for knowledge, glossary, etc.
- Clear logical structure of issues → measures → results
In other words, GEO means optimizing the structure, style, and information structure to be “easy for an AI to read and cite as a source.
This GEO perspective is not just about search inflow; it is a strategic element directly related to corporate recognition, visualization of expertise, and branding via AI. Particularly for BtoB companies, reliability and track record are the source of value, so visualizing that value as knowledge on the Web can be said to be at the core of corporate activities.
With the rise of generative AI, the role of a corporate website is no longer enough to be “search friendly. From now on, it will be important to be “chosen by AI. To achieve this, strategic use of the Web is required to accumulate the company’s knowledge, achievements, and verbalized value as organized primary information.
Metaphor provides content design support with an eye toward GEO and Large Language Model Optimization (LLMO). If you are considering creating a corporate website that is resistant to generative AI, please contact us.
