A Business Can Rank on Google and Still Disappear from AI Search

A company can spend years building a strong position on Google and still disappear when a potential client asks ChatGPT, Gemini or Perplexity whom to hire. A German study published in September 2026 tested 10’000 searches across ten sectors and found that 76.9 percent of businesses appearing in Google’s top ten were absent from the corresponding AI answers. AI assistants returned fewer than five businesses on average.

The research covers Germany, not Switzerland, so the 76.9 percent figure does not describe the Swiss market. It does expose a gap Swiss companies need to measure separately. Google tells a business where it ranks in search results. An AI assistant decides whether to include the business in its answer.

Someone searching for a web agency in Geneva might type “web agency Geneva” into Google and compare the names on the first page. The same person might ask ChatGPT, “Which Geneva agencies build multilingual B2B websites?” Instead of presenting a page of options, the system selects a handful of firms and leaves the rest out. A fourth-place Google ranking tells the company very little about whether it will make that shortlist.

The German research also found that many businesses recommended by AI assistants sat outside Google’s top ten. The researchers traced information to company websites, directories and specialist platforms. Swiss businesses therefore face a wider visibility test than their search ranking alone.

AI systems need enough information to identify what a business does, where it operates and which client need it matches. A Geneva agency that states it builds multilingual corporate websites for Swiss and international companies gives the system a clear service, location and client group. “Digital experiences for ambitious brands” gives it a slogan.

Websites often scatter the useful information. A company describes website development on one page, multilingual content on another and its Geneva office in the footer. A visitor browsing the site joins those details together. An AI assistant answering “Which Geneva agencies build multilingual B2B websites?” also needs to connect the service, language capability and location.

Internal terminology creates another issue. Clients search for a CRM consultant, multilingual website agency or SEO partner. The supplier calls the same service “Customer Experience Transformation” or “Digital Growth Solutions”. Unless the website also names the service in the language clients use, the brand phrase gives an AI system little evidence about what the company actually sells.

Information elsewhere on the web adds another layer. One directory lists the legal company name, another carries an old address, an industry platform describes services the business stopped offering and the website presents the latest positioning. People often recognise that all four references belong to the same organisation. AI systems still have to reconcile the differences.

Company name, location, services, sector and contact details need to agree across the sources where the business appears. Old profiles deserve attention because they keep outdated information in circulation long after a website redesign or repositioning.

The German report gives directories and specialist platforms a prominent role in AI visibility. The company behind the research also operates major German directory brands including Gelbe Seiten, so its claims about directory data need to be read in that context. The underlying finding is more useful than the sales message: businesses recommended by AI did not simply mirror Google’s first page.

External coverage gives AI systems more information about a company’s expertise. A consultancy describing itself as a specialist provides one source. A trade publication discussing its work, an industry association listing its expertise or a specialist platform placing it in the right category provides another. Firms with narrow specialisms benefit when several independent sources describe them in the same concrete terms.

Consider a fiduciary firm in Lausanne. Its website ranks well for the company name and broad local searches. A prospective client asks an AI assistant for “a Lausanne fiduciary experienced with international founders relocating to Switzerland”. The prompt combines location, profession, client type and service experience.

A generic page listing accounting, tax and advisory services gives limited evidence for the final two elements. Pages covering relocation work, founder clients and the relevant tax or corporate services give the system a clearer match. External references connecting the firm with the same work reinforce the association.

Switzerland adds a language problem. A Geneva business publishes in French and English. A Zurich company works in German and English. National firms operate across German, French and Italian markets. The web footprint often looks very different from one language to another because service pages, directory listings, media coverage and customer references do not appear evenly across all four.

Swiss companies need to test German, French, Italian and English queries separately. A firm appearing regularly in German-language answers might be absent from a comparable French prompt because the French-language web contains less information about its services.

SEO reports still show rankings, impressions, clicks, organic traffic and conversions. They do not show whether ChatGPT, Gemini or Perplexity names the company when someone asks for a recommendation. Searching the company name proves only that the system recognises it. A more useful test starts with the client’s need and leaves the company name out.

A digital agency might test “multilingual website development in Geneva”, “SEO agency for Swiss B2B companies” and “WordPress support in French-speaking Switzerland”. A hotel might test family accommodation near a destination or hotels with meeting facilities. An accountant might use the problems that bring clients through the door rather than the formal labels used on the services page.

Run the same prompts across several AI assistants and record who appears. Look at which competitors receive recommendations, how the systems describe them and which facts they get wrong. One firm might appear under an old service category. Another might lose out to competitors whose websites explain their work more precisely. A third might have a strong website but almost no relevant presence beyond its own domain.

Those results give marketing teams specific work. Rewrite vague service copy around the terms clients use. Correct inconsistent directory profiles. Remove obsolete addresses and services. Strengthen pages covering specialist work. Build credible third-party references through industry associations, partnerships, expert contributions and media coverage where the business has something substantive to say.

Technical teams have a different job. Search engines and AI retrieval systems need accessible pages, coherent site architecture and machine-readable information. Structured data helps identify organisations, addresses, services and other business details. It does not rescue a page that never states clearly what the company offers.

Google remains a major source of customer discovery, and SEO remains part of the job. The German study shows why a first-page ranking no longer gives management a complete picture of digital visibility. A business can sit near the top of Google and still be absent when an AI assistant gives a prospective client three names.

What to Do If Your Business Is Missing from AI Search

Check the searches that matter commercially

Test ChatGPT, Gemini and Perplexity with the questions a real client would ask. Do not include your company name. Use service, location, sector and client need.

Compare who appears instead

Record which competitors appear and how the assistant describes them. Look for differences in service clarity, location information, specialist pages and external references.

Make the offer easier to understand

State what you do, who you do it for and where you operate. Replace vague brand language where it hides the service people actually search for.

Fix inconsistent business information

Check your website, directories, industry platforms and company profiles. Remove old addresses, outdated services and conflicting descriptions.

Strengthen specialist pages

A broad “Services” page often gives too little context. Build pages around the work clients actually ask for, especially where location, industry or client type matters.

Check multilingual visibility separately

Test the same commercial query in German, French, Italian and English where relevant. A strong footprint in one language does not guarantee the same result in another.

Repeat the test

AI answers change. Recheck the same query set after major website updates, positioning changes or new third-party coverage.