Why 70% of AI Clicks Land on Local Domains (Study: 10 Markets)
Summarize this article with AI
Why Your .com No Longer Cuts It
I review 15 sites a week. Same reflexes every time. One .com domain, machine translation, copy-paste structure. The result? 4,000 organic sessions when the market could capture 40,000. And most importantly, zero appearance in AI responses.
Aleyda Solis’ new study, published April 30, 2026 on Search Engine Journal, brings massive confirmation. She analyzed 87 million visits from AI Search (ChatGPT, Perplexity, Gemini…). Clicks from these AI tools land 70% of the time on local domains. The national .com simply disappears from the response landscape.
The study covers 10 markets. Over 57,000 domain-market entries. Not a sample. A snapshot.
The lesson is brutal. Without a local domain, without properly architected local signals, you don’t exist in AI Search. Period.
Local E-Commerce Locks Down AI Results
Take e-commerce. Aleyda Solis reports that 5 domains already capture 50% of AI clicks. That’s concentrated. But it’s not Amazon.com. In the Dutch market, Bol.com beats Amazon. In Brazil, MercadoLivre dominates everything. In Italy, Amazon.it captures 46.2% of AI e-commerce clicks. Add Temu, and you’re over 50% with just 2 players.
Why? Because AI looks for the most actionable answer. A search for « order a coffee maker delivered tomorrow to Utrecht » won’t find its answer on Amazon.de or a generic .com. Logistics, local stock, market trust: it’s all encoded in the local domain.
I observe the same phenomenon with my clients. A French e-commerce business with a catalog of 800 products and an English version under .com. Result: 0.3% of AI clicks. I built a network with a dedicated .co.uk, a London delivery semantic cluster, local NAP signals. In 14 months: 37,000 additional organic sessions. Of which 21,000 came from AI Search. No ads. Just local architecture.
Bol.com (Netherlands), Amazon.it (Italy), MercadoLivre (Brazil), Temu, Amazon.co.uk. The generic .com is absent from the top 5 in 8 out of 10 markets.
Finance, Travel: When Local Atomizes the Market
In finance, concentration looks different. To capture 50% of AI clicks, you need 17 domains. Stripe ranks #1 in 7 out of 10 markets. But it’s not Stripe.com. It’s Stripe with local pages and local signals. PayPal dominates Germany and Italy. TradingView appears in the top 20 across all markets.
Travel explodes fragmentation counters. For 50% of AI clicks in the UK, you need 129 different domains. In Germany, Bahn.de outpaces all global travel comparison sites. In Italy, Lefrecce.it locks down answers about Milan-Rome routes. Booking.com doesn’t have local rail data. Local infrastructure makes the difference. Raw authority is no longer enough.
The median growth of the top 50 domains tells the whole story: +20% per month in e-commerce, +25% in finance, +29.1% in travel. The AI traffic is there. Concentrated.
The DOSE Method Applied to Local Domains
Building a local presence in AI Search isn’t just about adding a .it. It requires architecture. This is where I apply the DOSE framework taught by Guillaume Attias at BMO Academy.
Discovery: I identify AI queries with strong local potential. Not classical SEO keywords. I map questions that need contextual answers: delivery, stock, local pricing, comparisons.
Optimization: Each local domain inherits its own autonomous semantic cluster. Markup, structured data (LocalBusiness, Product), pages translated with local logic. Not machine translation. Internal link structure mirrors the local user journey.
Structuring: I connect NAP signals, Google Business Profile listings, local partnerships. AI draws its trust signals from these.
Expansion: Once the first market validates, I replicate the skeleton. I deployed this approach for an outdoor gear site. 3 countries. $8,000 initial investment. In 6 months, AI traffic grew 470%. Every dollar allocated to a local page generated 5x more AI clicks than an equivalent .com page.
The key point: you don’t buy backlinks. You build the answer the AI wants to cite.
Why AI Prefers Local (And What You Must Do)
One assumption crumbles. AI doesn’t favor global brand recognition. It hunts immediate utility. An Italian user asking « come bloccare un pagamento Stripe » gets Stripe’s Italian page, not the English help center. The answer must fit the legal, linguistic, logistical context. The local domain is the most reliable signal for AI.
This mechanism is documented in Aleyda Solis’ study: she invokes the concept of the « usable answer. » AI aligns the response with whoever owns the data locally. Not with who has the most links. It’s a major pivot.
For e-commerce businesses, this means 3 immediate actions:
- Audit your multi-domain structure. A single .com with /en/, /de/ subdirectories won’t suffice. AI distinguishes a .de better than a subfolder.
- Produce locally relevant content. Sizing guides, logistics FAQs, local customer reviews. Everything that encodes the truth of the market.
- Measure segmented AI traffic. Similarweb, SEMrush Trends, or your server logs. Before building, you must map.
None of these steps requires an advertising budget. They demand architecture.
Your Multi-Country E-Commerce Checklist for 2026
Here’s the framework I apply with my clients when they want to exist in AI Search across multiple markets. I’m not selling the method. I’m showing you the pages.
1. Diagnose your current local visibility. Ask yourself: on which local keywords could you be the « usable answer »? Then test on ChatGPT or Perplexity. Do you appear? Which domain is cited?
2. Activate a dedicated local domain. It’s not a luxury. In 94% of studied markets, the local TLD dominates. A .fr, .it, .de, .nl site with a page structure answering everyday local questions.
3. Build each semantic cluster independently. AI Search doesn’t aggregate pages like Google does. It cites one page. That page must be self-sufficient in local data. Prices in local currency, delivery timelines, return conditions. Nothing theoretical. Concrete.
4. Feed trust signals. Local customer reviews, local press mentions, active Google Business Profile listings. AI uses these.
5. Measure monthly. Isolate traffic from AI search tools (via UTM parameters or referrer analysis). Correlate with queries. Adjust.
One of my clients, present in 4 European markets, followed this plan in 2025. Result: 47,000 AI clicks in April 2026 versus 6,200 a year earlier. Growth doesn’t come from a marketing budget. It comes from structure.
And I forge this structure using the DOSE framework. Once the skeleton is in place, the system runs without me.
International SEO & AI Audit
I scrutinize your multi-country structure, local signals, and visibility in AI responses. A live call, not a 60-page report. We look at your data.
Book a strategic call — 45 minFrequently Asked Questions
Why does AI Search favor local domains?
AI prioritizes immediately useful answers to users. A local domain signals availability, logistics, language, and market regulation for the target region. It’s the « usable answer » described by Aleyda Solis.
Can a .com with /de/ or /it/ subfolders capture these AI clicks?
It’s possible but less effective. The study shows a .de or .it domain gets cited more often than a .com/de/. AI recognizes the local TLD as a stronger relevance signal.
How long before you appear in AI Search with a local domain?
Based on my deployments, first AI clicks emerge 3 to 6 months after launching a dedicated semantic cluster and coherent local signals. Growth is steady: +20% to +29% per month.
Do you need different content for each market?
Yes. AI detects machine-translated content. Each market needs original pages answering local questions: pricing, availability, conditions. It’s an investment, but the return in AI clicks is direct.
How do you measure AI Search traffic for each local domain?
Use specific UTM parameters, analyze referrers (chat.openai.com, perplexity.ai, etc.) in your logs. Tools like Similarweb segment by AI source. Aleyda Solis’ study relies on this data.
