Purchase decisions are increasingly starting not with Google, but with ChatGPT, Gemini & Co. And what the AI says about your company depends primarily on one source: your customer reviews. This article summarizes the most important findings from recent studies by McKinsey & Company (2025) as well as Kevin Indig, Citation Labs & Clickstream Solutions (2026) — and shows what workshops, practices, hotels, retailers, and hair salons should do now.
At a glance
• Around half of all Google searches already contain an AI summary; by 2028, this is expected to be over 75 percent.
• 74% of final purchase shortlists come directly from the AI response, without further verification.
• Over 90% of the sources for an AI response are third-party sources; your own website accounts for only 5 to 10%.
• Reviews and independent content make up more than 65% of cited sources.
• If you are not on the AI shortlist, you do not exist for these customers — there is no second page.
50%
of consumers are already actively using AI search — across all age groups
74%
of final purchase shortlists come directly from the AI response — without further verification
64%
of users do not click on a single website during their entire research process
The new gateway to the internet
Anyone looking for a restaurant, a workshop, a doctor, or a hotel today is increasingly asking an AI first — and instead of ten blue links, they receive a ready-made recommendation with two or three names. McKinsey therefore calls AI search the "new gateway to the internet."
The pace of this shift is remarkable: already, about half of all Google searches contain an AI summary — by 2028, this is expected to exceed 75 percent. By then, according to McKinsey, AI search will influence consumer spending of approximately 750 billion US dollars.
Share of Google search queries with AI summaries
Today
2028 (Forecast)
Source: McKinsey & Company, 2025
Experts call the new discipline behind this GEO — Generative Engine Optimization: the optimization for being found and recommended by AI systems.
How the AI knows what to say about your business
AI assistants answer questions based on trained knowledge and by actively looking up current sources. In both cases, the same applies: Your own website plays hardly any role. Over 90 percent of the sources that AI responses rely on are third-party sources.
What AI responses are based on
Reviews & independent content
Other third-party sources (directories, trade, etc.)
Company's own website
Shares of sources cited by AI systems, rounded. Source: McKinsey & Company, 2025
What an AI says about your company therefore depends on what trustworthy external sources say about you — above all, genuine customer reviews. An AI does not "believe" your advertising: it compares many independent voices with one another. For an AI, reviews are what word-of-mouth is to your regular customers — the most credible evidence of quality.
The AI shortlist decides
A behavioral study with 48 participants and 185 documented purchase tasks — ranging from laptops to insurance — shows how strongly AI recommendations are already steering purchasing behavior (Kevin Indig, Citation Labs & Clickstream Solutions, April 2026):
- 74% of final shortlists were adopted directly from the AI response — without further verification.
- 64% of searches were conducted without a single website click.
- 93% of purchase decisions in the laptop example were limited to just three recommended brands.
„I don't know the provider, but the AI says they are the cheapest. That's enough for me."
— Participant in the behavioral study regarding an unfamiliar provider
Anyone who is not on the AI list simply does not exist for these customers. There is no page 2 where one might still be found — the shortlist is the decision.
Reviews: Your strongest lever
Both studies lead to the same conclusion: customer reviews are the most effective lever you have under your control. Four measures with the greatest impact:
- Continuously collect new reviews. AI systems prefer information that is confirmed frequently and currently. A steady stream of new reviews signals: This company is active and relevant.
- Respond to every review. Responses generate additional, consistent first-hand content and show that feedback is taken seriously.
- Be present at all locations. The AI compares many sources — every location without fresh reviews is a gap in the overall picture of your company.
- Keep an eye on developments. Only 16% of companies systematically monitor their AI visibility — those who measure gain the advantage.
Self-test: Does the AI recommend your business?
Ask ChatGPT or Gemini the way your customers would — e.g., "Best hairdresser in [Your City]?". Are you mentioned? With what description? The result shows you where you stand today.
Collect more reviews now →Frequently Asked Questions (FAQ)
What is GEO (Generative Engine Optimization)?
GEO is the optimization of a company for being found, cited, and recommended by AI systems like ChatGPT, Gemini, or Perplexity — analogous to SEO for traditional search engines.
Why are reviews so important for AI search?
In many industries, over 65% of the sources that AI responses rely on are reviews, testimonials, and independent content. In contrast, your own website accounts for only 5–10% (McKinsey, 2025).
How do I find out if ChatGPT recommends my company?
Ask an AI the question your customers would ask, e.g., "Best car repair shop in [City]?". If your business is not mentioned, the AI lacks trustworthy external signals — especially current reviews.
What can I do specifically as a small business?
Systematically collect reviews (e.g., via NFC chip or QR code directly at the location), respond to every review, maintain all locations, and regularly measure development. That is exactly what shoptimizer was developed for.
Sources: McKinsey & Company: "New front door to the internet — Winning in the age of AI search", October 2025 · Kevin Indig, Citation Labs & Clickstream Solutions: Behavioral study on purchase decisions with AI assistants, April 2026 (48 participants, 185 purchase tasks) · futurezone.at / Alexander Rus (Evergreen Media): "How does ChatGPT know what it says about a brand?", 2026.
AI tools were used for this article. Content and sources were edited and verified.
