Leader
43Heralded Score, range 36–53
The most visible conversational AI platforms in Germany, October 2026
Most cited source
6citations
From Google AI Overviews
Where engines disagree most
1/4engines put it in their top three
Top three on Gemini, not on Google AI Mode, Google AI Overviews and ChatGPT
Brands
Search is not measured for this board yet, so it ranks on AI answers.
| Rank | Brand | Rank change | Heralded Score All | AI | Score trend | Mentioned | Recommended | By engine | Mentioned by | Change |
|---|---|---|---|---|---|---|---|---|---|---|
| – | | Unranked | Too few answers | Not measured | –
| 3% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 1% | 0% | | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 2% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 0% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 3% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 1% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 3% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 3% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 0% | 0% | None | First reading | |
| – | | Unranked | Too few answers | Not measured | –
| 2% | 0% | None | First reading |
“=” marks brands whose score ranges overlap, so the test cannot separate them. Every brand named in at least three answers is ranked. AI gives each brand’s score and rank from the engines’ answers; Search is not measured for this board yet. By engine has one dot per engine (ChatGPT, Google AI Overviews, Gemini and Google AI Mode), darker the more often it names the brand.
Most cited sources
| Rank | Source | Type | Cited in | Engines | Pages cited | Brands its pages mention |
|---|---|---|---|---|---|---|
| 1 | | Social | 6.7% | 1 of 4 | 6 | Not read |
| 2 | | Social | 1.7% | 1 of 4 | 5 | Not read |
| 3 | | Vendor site | 3.3% | 2 of 4 | 4 | Not read |
| 4 | | Vendor site | 3.3% | 1 of 4 | 4 | Not read |
| 5 | | Vendor site | 5.0% | 3 of 4 | 3 | Not read |
| 6 | | Regulator | 1.7% | 1 of 4 | 3 | Not read |
| 7 | | Industry authority | 3.3% | 1 of 4 | 3 | Not read |
| 8 | | Vendor site | 1.7% | 1 of 4 | 3 | Not read |
| 9 | | Vendor site | 5.0% | 3 of 4 | 2 | 1 |
| 10 | | Community | 5.0% | 1 of 4 | 3 | Not read |
Questions asked
| Question | Language | Most recommended |
|---|---|---|
| Was kann ein Support-Team nutzen, um Kundenfragen per KI-Chat zu beantworten? | German | |
| Welche Plattform hilft einem Unternehmen dabei, einen Chatbot für Kundengespräche aufzubauen? | German | |
| Was sollte ein Onlineshop nutzen, um Kundenchats automatisch zu bearbeiten? | German | AVA Book and |
| Wie kann ein Unternehmen KI-Chat in seinen Kundenservice integrieren? | German | No brand recommended |
| Wie baue ich einen Chatbot, der Fragen zu unseren Produkten beantworten kann? | German | |
| Wie können wir verhindern, dass Kundenchats außerhalb der Geschäftszeiten unbeantwortet bleiben? | German | No brand recommended |
| Wie bringe ich einem KI-Chatbot bei, Fragen anhand unseres Hilfe-Centers zu beantworten? | German | No brand recommended |
| Wie kann ein Support-Team komplexe Chats vom Bot an einen Mitarbeitenden übergeben? | German | |
| Sollten wir einen KI-Chatbot für Vertrieb und Support nutzen oder getrennte Bots? | German | No brand recommended |
| Was funktioniert im Kundenservice besser: ein KI-Chatbot oder ein skriptbasierter Bot? | German | No brand recommended |
Each engine's top three
Google AI Mode
Gemini
- 1
Zendesk AI
- = 2
Botpress
- = 2
Fin AI Agent
Google AI Overviews
- 1
- = 2
- = 2
HubSpot Service Hub
ChatGPT
- 1
- 2
- 3
Tidio
Related leaderboards
Method
Each leaderboard ranks brands on two sides. AI is how ChatGPT, Gemini, Google AI Overviews and Google AI Mode answer the board's buyer questions. Search is Google's top 10 results for the keywords behind those questions. All combines the two, where Search is measured.
For the AI side, each month Heralded asks every engine the same buyer questions, twice each, and reads every answer. An answer counts 0 for a brand it leaves out, 50 for a brand it names and 100 for a brand it recommends, and the Heralded Score is built from those. A brand named in at least three answers gets a score and a place.