Methodology

Which sites we checked

The list comes from three public rankings with no selection of our own: 215 sites in total. For each we checked the site German visitors land on; for international brands, the German storefront.

• DAX 40 constituents after the September 2026 index review (STOXX, Deutsche Börse)
• EHI Retail Institute, Top 100 online shops in Germany, 2025 edition (fiscal 2024)
• agma DNA, digital offerings ranking, August 2026, ranks 1 to 79 (at least 1% reach)

Agent readiness

23 checks in five areas: discoverability, content accessibility, bot access control, APIs with auth and MCP, and agentic commerce. They follow the published specifications (among them RFC 9309, RFC 8288, RFC 9727, llms.txt, MCP, A2A, UCP and ACP). Levels 0 to 5 follow the same rules as Cloudflare's isitagentready.com; commerce protocols do not count toward the level.

The 0 to 100 score weights the areas 25, 20, 22 and 33 percent. Partly met checks earn partial credit. When a firewall blocks our request or a server does not answer, the check counts as could not check, never as not met. Our scanner identifies itself openly as RezolveAi-AgentReadiness/1.0. Sites that turn such requests away were checked again through a fetching service with a German IP address, and the block itself stays a finding. Response headers can't be read that way, so the two checks that need them count as could not check there. As of 21 September 2026.

Visibility in AI answers

For each brand we wrote 15 brand-neutral questions in German, the way people in Germany ask about its category. No question names the brand or its competitors. The questions went through Rezolve Ai's AEO platform to: OpenAI GPT-5 mini with web search (API); Anthropic Claude Haiku 4.5 with web search (API); Perplexity Sonar (API); Google Gemini, queried like the app. Every engine was queried with the region Germany. None of them accepts a language setting through the platform, and the questions themselves are German. Measured from 19 September 2026 to 21 September 2026.

The mention rate is the share of answers that name the brand. Sentiment and citations come from the analysis of each answer. The OpenAI, Claude and Perplexity models were queried through their APIs with web search; their answers can differ from the consumer apps. Gemini is queried like the app.

AEO score

The AEO score combines visibility (32 percent), citations (22), sentiment (18) and agent readiness (10). When a value is missing, the others are reweighted. Because every brand gets its own questions about its category, we compare visibility and the AEO score only within an industry. The overall ranking sorts by agent readiness, the one test that is identical for every site.

Statistics

Industry values are medians with the middle half. We measure relationships with Spearman's rank correlation and differences between the three sources with the Kruskal-Wallis test. We report only three relationships fixed in advance, each with sample size and p-value. An association is not proof of cause and effect.

Limits

The results are a snapshot. AI answers change constantly and websites keep evolving. We recognise brands by name, so unusual spellings can be missed. Quoted AI answers are unverified and shortened; we do not quote critical mentions.

Corrections

If a result looks wrong to you, contact us via rezolve.com/contact-us. We review every report.