There is no publicly accessible, fixed ranking for AI-generated sources . Response systems evaluate, based on the question, the system itself, and the available information, which sources are thematically relevant, understandable, contain reliable data, and appear sufficiently up-to-date. This leads to a clear task for your SME: Don't rely on supposed AI tricks, but rather on precise offers, verifiable statements, and consistent company information.
This is not a ranking guarantee and not a promise to be mentioned in a response from ChatGPT, Google, or Perplexity. However, it does lay the foundation for humans, search engines, and AI applications to correctly understand your business. Google continues to describe the same basic requirements for AI Overviews and AI Mode: technically accessible pages, helpful and reliable content, and adherence to search guidelines. Google does not specify any additional special requirements for these.
AI answers don't arise from a single signal. The quality, clarity, and verifiability of the information a system can find in response to a question are crucial.
In my work with small businesses, I often see the same bottleneck: The expertise is there, but the services offered are described too generally on the website, contact details differ between channels, or important claims remain unsubstantiated. Before any further content, technical enhancements, or campaigns are developed, this informational foundation needs to be sound.
How AI source selection works from question to answer
The individual systems operate differently and do not fully disclose their selection models. Nevertheless, the basic process of AI source selection can be described in an understandable way. It is not about a secret algorithm, but rather about several checks that can change depending on the user's query.
- The user question is categorized. The system recognizes the topic, intention, language, location, and expected level of detail in the response. The question "Who designs a hotel website in South Tyrol?" requires different information than "What does a corporate design cost?".
- Information needs and entities are identified. Entities are clearly identifiable units such as companies, people, services, places, products, or organizations. The system attempts to understand who offers something, where the service is provided, and what relationships exist.
- Relevant content is retrieved. This step is called retrieval: Relevant text passages, pages and sources are searched for from available indexes, data sources or web content.
- Sources and statements are evaluated. Factors such as thematic relevance, comprehensibility, originality, topicality, technical accessibility, and context of trust can play a role.
- The answer is formulated. In the answer synthesis, the system summarizes the information found, categorizes it, and provides sources or links depending on the product and situation.
The crucial point is this: depending on the system and product, an answer doesn't necessarily access the web live, display sources, or retrieve every relevant website. Therefore, mentioning a source in AI-generated answers isn't controllable like an advertisement. However, you can control whether your company offers a clear, credible, and technically feasible source of information.
The seven factors for AI sources in a self-assessment
The following seven factors are not a purported AI algorithm. They are a practical testing model for your website and your public information base. For an SME, I would approach them in this order: first clarity, then evidence, then external confirmation.
1. Thematic relevance: Does your page answer a specific question?
What it means: Thematic fit occurs when user question, service offering, and actual experience align precisely. A page about "digital solutions" is of little help if someone is specifically looking for a website for a craft business, brand positioning, or AI-supported process simplification.
Here's how to check: Can an interested person understand after just a few sentences who your offer is for, what problem you solve, what specific service you provide, and in what context you have experience?
Your action: Create or revise service pages based on real questions and decisions. Don't just describe activities, but also the initial situation, the approach, the result, and the limitations. A clear website strategy connects the logic of the offer, user guidance, and content, rather than optimizing individual texts in isolation.
What this doesn't promise: A precise performance page doesn't guarantee a mention. However, it reduces the risk that your offer will be categorized too generally, incorrectly, or not at all.
2. Entities: Is your company clearly identifiable?
What it means: Entities help systems correctly identify names and relationships. These include company names, responsible individuals, services, locations, target audiences, brands, and official profiles. This clarity prevents confusion, especially with similar company names or regional offerings.
Here's how to check: Are the company name, address, phone number, responsibilities, and service descriptions the same on the website, Google business profile, industry profiles, and social networks? Is it clear who is behind the company?
Your action: Create a binding master data list and check all official profiles. Add a clear company page, contact persons, and verifiable service relationships to the website. Structured data can further describe this information in a machine-readable format. However, it does not replace an understandable website and consistent information.
What this doesn't promise: Clear data doesn't automatically lead to a recommendation. However, it does create the conditions that a mention will actually be attributed to your company.
3. Original information: Is there anything in your work that is not interchangeable?
What it means: Original information consists of original, verifiable contributions: documented processes, local experience, concrete case studies, proprietary methods, measurements, photos from real projects, or reasoned assessments. Generalities that any company could formulate hardly create any substance.
Here's how to check: Could a competitor copy the same text almost unchanged onto their website? If so, you probably lack personal experience or a clear perspective.
Your action: Document typical questions from consulting and project work. Explain which decision you recommend and why, what prerequisites apply, and which errors you repeatedly observe. A good case study doesn't need to disclose confidential figures. However, it should clearly describe the initial situation, the approach taken, and a verifiable result.
What this doesn't promise: Your own content won't automatically become AI answer sources. However, it provides more citable substance than reworded standard texts.
4. Source evidence: Can you provide evidence for important statements?
What it means: Sources make external facts verifiable. This applies to legal requirements, studies, technical claims, market information, or statements about platforms. An assessment based on your experience can remain as an assessment, but should not be presented as a universally valid fact.
Here's how to check: Does your text contain specific figures, legal statements, or promises for which no primary source, study, or reputable expert source is cited? Are the links still accessible?
Your action: When citing external facts, link to the original source whenever possible. Include publication or update dates if content is subject to change. Remove unsubstantiated claims or clearly label them as experience, assessment, or example.
What this doesn't promise: A link to a source is not a guarantee of trust in the entire website. However, it does make individual statements verifiable and protects you from unnecessary exaggeration.
5. Up-to-dateness: Is your information still accurate?
What this means: Up-to-dateness is not the same as constantly publishing new blog posts. What matters is whether services, contact persons, pricing information, legal notices, product information, and sources are still valid.
Here's how to check: Are there pages with outdated dates, discontinued services, former team members, or links to obsolete information? Do you know when an important page was last reviewed by a professional?
Your action: Implement a simple editorial cycle: Review core pages quarterly, update time-sensitive information immediately when changes occur, and evaluate technical articles at least once a year. Update not only the date, but also the content, references, and examples.
What this doesn't promise: A new date is not a sign of quality if the content remains outdated. The content itself must be clearly up-to-date.
6. Consistency: Do your channels tell the same story?
What this means: Consistency concerns core data and key messages. Your company doesn't need to communicate word for word everywhere. However, name, location, value proposition, responsibilities, and contact options should not contradict each other.
Here's how to check: Does your homepage have a different service description than your industry profile? Is a location spelled differently? Are team details and images still correct?
Your action: Define a central source for company data and a concise brand description that all stakeholders use. Especially when the website, sales department, social media, and external service providers are involved, this foundation prevents information chaos. This is a core component of a good brand strategy : The brand must clearly communicate what it stands for.
What this doesn't promise: Consistency is no substitute for quality. Above all, it prevents trust from being lost through inconsistencies.
7. External validation: Is your expertise also visible outside of your website?
What it means: External validation comes from reputable industry profiles, relevant mentions, local institutions, editorial contributions, partnerships, or verifiable references. These signals place your company in a broader context.
Here's how to check: Are there independent bodies that can meaningfully assess your company, your expertise, or your projects? Are these references genuine, current, and thematically relevant?
Your action: Maintain official industry and location profiles, request honest feedback after completed projects, and document collaborations transparently. Contextual brand mentions can be valuable for local businesses; you can find more information in the article about local co-citations for SMEs.
What this doesn't promise: External validation must not be bought or artificially generated. Purchased reviews, bought links, and fabricated testimonials are not sustainable signals of trust. The European Commission defines disinformation as demonstrably false or misleading information that is disseminated, among other things, for the purpose of deliberate deception or economic gain and that can cause public harm.
Self-assessment for SMEs: seven yes/no questions
You don't need a major audit to get started. Answer these seven questions honestly with yes or no:
- Does each core service have its own page that clearly explains the target group, problem, approach and result?
- Are the company name, contact person, location, and contact details identical across all official channels?
- Do your most important pages contain original information from your actual work?
- Are external facts supported by reliable source evidence?
- Have any of the core pages been reviewed for content within the last twelve months?
- Are performance promises, brand descriptions, and responsibilities consistent across all channels?
- Is there genuine, relevant external confirmation instead of bought or manipulated signals?
Don't prioritize based on the number of "no" responses, but rather on impact. A lack of clarity in your offerings and conflicting entities will block all further action. Only after addressing these issues will additional content, structured data, or technical improvements be worthwhile.
A realistic 30-day start plan
Days 1 to 10: Organizing supply and entities
- Check the homepage, contact page, and three main service pages.
- For each task, formulate a clear answer to: For whom, what problem, what approach, and what result?
- Matches company name, location, phone number, email address and contact person with official profiles.
Days 11 to 20: Follow up with evidence and updates
- Highlight unsubstantiated external claims and add or remove reliable sources.
- Replace general texts with at least one concrete example from your experience.
- Check for outdated content, images, team information, and links.
Days 21 to 30: Strengthening the context of trust
- Update relevant industry and location profiles.
- Please ask selected partners or clients for honest, specific feedback.
- Schedule a recurring appointment to maintain your core information.
When information is scattered across different documents, systems, and minds, more content isn't the solution. Your company first needs a shared information base. Our AI and digitalization services address this need: organizing processes and knowledge so that information remains reliably accessible.
Questions on the selection of AI answer sources
Can my company rank in ChatGPT or other AI systems?
The term "rank" is only partially appropriate because answer systems don't have a unified, publicly visible results list like a traditional search engine. However, you can improve the conditions so that your company can be found and correctly categorized with clear, up-to-date, and verifiable information. A mention in the results doesn't guarantee a ranking.
Does every page need structured data?
No. Structured data is useful when it precisely describes specific information such as companies, services, people, events, or products. It helps with classification, but it doesn't replace good content or a technically sound website. Start with the core pages that contain important company data and offers.
Are good reviews sufficient as a sign of trust?
Reviews can be helpful if they are genuine, specific, and verifiable. However, they are not enough on their own: a website still needs clear performance data, consistent information, original details, and up-to-date contact information. Purchased or artificially generated reviews do more harm than good in the long run.
How often should I update website content?
Always update content whenever services, responsibilities, pricing, sources, or legal frameworks change. You should briefly review core pages at least once a quarter and specialist articles at least annually. A genuine content revision is crucial, not just a new date.
Can an AI mention be guaranteed?
No. Which sources a system accesses, how it prioritizes content, and whether it visibly cites sources depend on the question, the system, the time, and the available information. Reputable work improves information quality and discoverability, but does not guarantee a specific placement or mention.
First the information base, then the expansion
If your offerings, data, and evidence are inconsistent, don't rush into producing additional content or purchasing specialized technical solutions. First, clarify what your company offers, who represents it, which claims you can substantiate, and which information is up-to-date. This strengthens the foundation for AI-powered responses and, most importantly, builds trust with the people who will ultimately choose your company.
Visibility should not be based on manipulation. It should grow from competence, clear communication, and verifiable information. This is more complex than a short-term trick, but significantly more sustainable for small businesses.
