When the AI Recommends: The Guide to AI Visibility in South Tyrol and the DACH Region

You have probably done it yourself. Instead of searching Google and clicking through ten links, you asked an AI — "What's the best restaurant nearby?", "Who will reliably build me a new website?", "Which provider fits my budget?" — and got a finished recommendation with two or three concrete names. Without a single click on a website.

That is exactly what your customers are doing now. When someone asks about a business like yours, the AI names either your name — or your competitor's.

It is no longer just about being found, but about being recommended — by a machine that decides on its own whom to name. This guide puts the topic in context and goes deep where it matters most for South Tyrol. The individual technical and risk topics are already covered in detail elsewhere — we link onward from there.

The essentials in brief

• AI search is everyday life in the DACH region and Italy: 65% of Germans, around 60% of Swiss and over two thirds of Austrians use AI tools; in Italy, usage doubled within one year — to 35% of the online population (TÜV 2025, IGEM 2025, APA 2025, Audicom/ANSA 2025).
• But Italy's businesses are lagging behind: only 8% of Italian SMEs use AI, compared to 25% in Germany (Eurostat/Politecnico di Milano, IfM Bonn 2025). Customers are far ahead of their own economy — the spot in AI answers is still free.
• Your website is the most important AI source: 86% of all AI citations come from brand-owned channels; your own website leads with 44% (analysis of 6.8 million AI citations, 2025).
• Substance is measurably rewarded: statistics, sources and expert quotes in the text increase the chance of being cited by AI by up to 40% (GEO study, ACM KDD 2024).
• Fewer clicks, better customers: AI visitors convert significantly better and stay noticeably longer on the website than classic organic traffic (industry analyses 2025) → in detail in Fewer Clicks, Better Customers.
• Being cited wrongly is the bigger risk: AI search engines got source attributions wrong in over 60% of test cases (Tow Center, Columbia 2025) → in detail in When AI Cites You Wrong.

How do people in the DACH region and Italy search today?

The short answer: increasingly via AI — and in all four markets that matter for South Tyrol.

In Germany, according to the TÜV AI Study 2025 (Forsa, 1,005 respondents), 65% of the population uses generative AI; among 16-to-29-year-olds it is 91%. The most common use case is research: 72% of users search for information with AI — a year earlier it was just 48%.

In Switzerland, AI usage rose from 40% to 60% of the population within one year according to the IGEM Digimonitor 2025 — around 3.8 million people. Among 15-to-34-year-olds it is already 79%.

In Austria, according to APA's "AI Readiness Study 2025", more than two thirds of the population uses AI. Among young people, saturation is almost total: 94% of 11-to-17-year-olds use AI chatbots, almost half of them for searching and summarising information (ÖIAT/Saferinternet 2025).

And Italy? It is catching up at record speed. According to Audicom/Audiweb data, around 15 million Italians — 35% of the online population — used AI apps in December 2025; the annual average doubled compared to 2024 (from 5.2 to 12.7 million). ChatGPT alone reaches 10.1 million users, almost one in four Italians between 18 and 74. Among 15-to-24-year-olds, ChatGPT penetration is at 49% according to Comscore — the app has more users there than TikTok. And the EU Kids Online report sums it up with a quote Italian teenagers told the researchers: "ChatGPT è diventato il mio nuovo Google" — ChatGPT has become my new Google.

The trend also comes with a hard forecast: according to Statista, around 13 million people in the US used generative AI as their primary search tool in 2023 — by 2027 it is expected to be over 90 million. These customers — guests, business partners, applicants from the DACH region and from Italy — are asking the AI about businesses like yours today.

Why is this an opportunity for South Tyrol in particular?

Because the decisive fault line runs right through Italy itself — between customers and businesses.

On the demand side: AI-savvy customers from all four markets. 60–65% of the DACH population uses AI, and Italian consumers doubled their usage within a year — to 35% of the online population (see above). On the supply side: standstill. In Italy, according to Eurostat, only around 8.2% of companies use AI technologies — well below the EU average. The Osservatorio Artificial Intelligence of the Politecnico di Milano shows the gap even more sharply: 71% of large Italian companies have at least one active AI project, but only 8% of SMEs. For comparison: in Germany, one in four SMEs now uses AI (25%; in 2023 it was just 11%), above the EU average of 19% (IfM Bonn).

In one sentence to remember: customers — including Italian ones — are already searching via AI; the businesses you compete with are not yet findable there. Almost one in four Italians uses ChatGPT, but only one in twelve small Italian businesses even has AI on its radar. This gap is the rare case where a small business can gain a structural head start with manageable effort — in three language areas at once. The window will not stay open forever: user numbers in Italy double every year, and the businesses will follow. Anyone investing in South Tyrol now can also benefit from the provincial digitalization grant — more on that in 60% grant: South Tyrol's digitalization aid until 2028.

What is zero-click search — and is it ruining my business?

Definition: zero-click search means a search query ends without the user visiting a website — the answer comes directly from the search engine or AI. The trend is clear: when an AI overview appears, the zero-click rate rises above 80%; on pure AI search services it is around 93% without any website visit.

Sounds like doom — but it is not. The clicks that remain are worth more: AI visitors convert significantly better and stay noticeably longer on the website, because the AI has already pre-filtered. The only precondition is appearing in the recommendation at all. The full numbers analysis: Fewer Clicks, Better Customers.

What does "AI-ready" really mean? (And what it doesn't)

Definition: a website is AI-ready when the major AI systems — ChatGPT, Gemini, Perplexity — can read it without errors, classify it correctly and recommend it as a reliable source.

The most common misconception first: AI-ready does notmean putting a chatbot on your own site. It means building the website so that the AI systems out there understand it — how an AI technically breaks down and processes a page for this is explained in detail in From SEO to GEO.

And guessing is a documented risk: if you don't document your facts in an unambiguously machine-readable way, you leave your company image to chance. For a business, being cited wrongly is the underestimated, bigger risk — even before invisibility. The full research on this: When AI Cites You Wrong.

Whom does the AI cite — and what can I influence?

You control 86% of the sources the AI builds its answers from yourself, or can influence them directly — that is the most important and at the same time most encouraging finding for small businesses. This comes from a large analysis of 6.8 million AI citations across ChatGPT, Gemini and Perplexity (2025).

The distribution in detail:

• Your own website — 44% of AI citations — full control.
• Directories & listings — 42% of AI citations — directly influenceable.
• Reviews & social — 8% of AI citations — influenceable.
• Forums (e.g. Reddit), news, other — 6% of AI citations — little control.

Two nuances you should know. First, the models tick differently: Gemini prefers structured, brand-owned websites (52.1% of its citations), ChatGPT leans more on directories and listings (48.7%), Perplexity spreads across industry services and review portals. A well-maintained own website and consistent directory entries therefore both pay off. Second: Google AI overviews cite pages that already rank well classically in around 97% of cases (industry analysis 2025). Classic SEO is therefore not dead — it is the ticket in.

What makes content citable? The GEO formula

Definition: Generative Engine Optimization (GEO) is the practice of optimising content to be cited as a source in AI-generated answers — coined by Aggarwal et al. (Princeton, Georgia Tech and others), presented at ACM KDD 2024, with a measured effect of up to 40% more visibility.

The three most effective levers in brief: concrete statistics instead of qualitative claims, citing real sources, including expert quotes with name and role. Keyword repetition — the old SEO reflex — showed no positive effect. The rule of thumb for every text: write so that a single paragraph, standing alone, answers a question completely and verifiably.

The full technical implementation — chunking, embeddings, structured data, RAG — is covered in From SEO to GEO.

Why is multilingualism an AI factor?

Because AI models work across languages but need clean, language-specific signals for reliable citations — content that is only auto-translated counts as half a website to the AI. For South Tyrol there is an added layer missing from almost every AI guide: language. A guest asks the AI in Italian, a business partner in German, a tourist in English — and all three should land at your business.

AI models work across languages, but they need clean signals for it. The citable key sentence: an automatically translated website is, to the AI, half a website. If your German version is perfectly maintained but the Italian one is just a neglected translation without its own URLs, its own SEO and its own structure, you lose exactly where Italian-speaking customers ask. What matters is that every language has a fully-fledged version of its own — its own texts, its own addresses (slugs), its own metadata — and that the language versions are technically cleanly connected (hreflang). A missing translation should be made visible, not silently filled with the main language.

For a trilingual location like South Tyrol this is not a tiresome extra but a structural competitive advantage: anyone who maintains DE/IT/EN cleanly is citable in three language areas — most competitors manage at most one. How this quiet local advantage plays out in practice is shown in Three languages, one citability: South Tyrol's quiet advantage.

How do I stay visible without giving away my data?

The answer: open standards let you declare machine-readably what is allowed — visibility in AI search yes, use for model training no; as an EU business you do not have to choose between visibility and control. Being perfectly machine-readable, however, also makes you an attractive target for AI training. Visibility and data sovereignty are not mutually exclusive — you just have to build them together.

The full explanation incl. the legal context (W3C TDM Reservation Protocol, EU AI Act): Visible to AI, Protected from Training.

What technology does an AI-ready website need?

In short, the five building blocks without which the best content remains ineffective: structured data (Schema.org/JSON-LD), clean semantic HTML, performance & mobile (stable Core Web Vitals), crawling permissions for AI bots (correct robots.txt, current sitemap, plus llms.txt), and agent readiness (Google's Lighthouse category "Agentic Browsing" since May 2026, WebMCP as an emerging standard).

The technical depth is covered in From SEO to GEO.

Self-test: which stage is your business at?

Instead of an endless checklist, an honest maturity model in four stages:

Stage 0 — Invisible. Slow site, patchy or only automatically translated multilingualism, no structured data. Barely readable for AI. (This is where — see the 8% figure — a large share of local businesses stands.)

Stage 1 — Findable. Solid technology, good classic SEO, well-maintained Google business profile. You get found, but AI cites you unreliably or with outdated details.

Stage 2 — Citable. Structured data, self-contained answers and FAQ, clean multilingualism, visible authority (E-E-A-T: Experience, Expertise, Authoritativeness, Trustworthiness — visible authors with qualifications, real customer voices). AI understands your business and names it correctly — in all three languages.

Stage 3 — Agent-ready. Additionally machine-readable company profiles (llms.txt and relatives), a controlled interface for AI agents (MCP/WebMCP), granular data sovereignty (TDM opt-out). AI can not only recommend your business but interact with it on your terms.

The goal is not to jump to stage 3 overnight, but to reach the next stage systematically. Every stage is measurable progress — and even the step from 0 to 1 sets you apart from the local average.

How btlabs Core solves this

btlabs Core is built as a platform for exactly this path. All company data lives in one central, structured base (single source of truth) — exactly what the citation-source analysis above identifies as the most important AI source. From it, the following are generated automatically: Schema.org/JSON-LD, clean metadata, hreflang, llms.txt and an MCP endpoint for authorized agents. Every language (DE/IT/EN) is a fully-fledged version of its own with its own URLs and its own SEO. Delivery is fast, analytics cookieless, hosting a dedicated EU server, the AI training opt-out granularly controllable.

The platform thus covers stages 2 and 3 without you having to assemble every building block yourself.

And the honest limit that belongs to this guide: no system guarantees #1 on Google or a mention in every AI answer — not even btlabs Core. No one who promises that is credible. What a clean foundation delivers: it makes mentions significantly more likely and ensures that when you are named, you are named correctly — with the right opening hours, the right services, the right contact. The substance has to come from you. We make sure it arrives. Curious who "we" actually are behind this guide? More on that at Web agency Bolzano — the alternative.

Sources & studies

• TÜV AI Study 2025 (Germany) — TÜV-Verband / Forsa. tuev-verband.de
• IGEM Digimonitor 2025 (Switzerland) — 60% AI usage, +20 percentage points in one year.
• AI Readiness Study 2025 (Austria) — APA/Google Austria; additionally ÖIAT/Saferinternet 2025.
• Digital Decade Country Report Italy 2025 — Eurostat/EU Commission: 8.2% AI usage among Italian companies. digital-strategy.ec.europa.eu
• Audicom/Audiweb analysis (V. Cosenza), via ANSA — Italy: 15 million AI users, doubling within one year. ansa.it
• Osservatorio Artificial Intelligence — Politecnico di Milano: 71% large companies vs. 8% SMEs.
• Digitalization of SMEs in EU comparison — IfM Bonn: 25% AI usage among German SMEs. ifm-bonn.org
• GEO: Generative Engine Optimization — Aggarwal et al., ACM SIGKDD 2024: up to 40% visibility gain. arxiv.org/abs/2311.09735
• AI Search Has a Citation Problem — Tow Center, Columbia 2025: >60% wrong source attributions. cjr.org
• Lighthouse Agentic Browsing — Chrome for Developers, 2026. developer.chrome.com
• The /llms.txt proposal — Jeremy Howard, Answer.AI. llmstxt.org
• W3C TDM Reservation Protocol — W3C Community Group, 2024. w3.org/community/tdmrep
• Model Context Protocol (MCP) — Anthropic. modelcontextprotocol.io
• AI Citations Study — Yext Research, October 2025: 86% of AI citations from brand-owned sources, analysis of 6.8 million citations across ChatGPT, Gemini and Perplexity. yext.com

Note: GEO, llms.txt and WebMCP are young, evolving standards. Impact and adoption are growing, but no major AI provider formally guarantees them as a ranking signal. All figures reflect the state of the respective survey.