Tracking your AI visibility is good, acting on it is better: 10 practices that really shape AI answers

Monitoring tools show whether AI engines mention your brand. The real question comes next: what actually moves the answer?

At SEO & GEO Summit in March 2026, Julien Bismuth (Getfluence) and Olivier de Segonzac (Resoneo) answered with 10 practices they see influencing AI answers and brand mentions, backed by field tests, a close reading of ChatGPT's inner workings and real campaigns.

Their starting point: generative AI calculates rather than thinks. It predicts the next word by probability. A mechanical process can be understood, and what can be understood can be influenced.

In this replay:

• How to build a test prompt set when conversational queries have no search volume: personas built from keyword data, synthetic prompts that trigger search, and brand persistence measured over 100 runs of the same question.
• Why ChatGPT leans so heavily on Google's index, why 50% of its fan-outs are written in English, and why meta descriptions and URL slugs matter again (the snippet comes from the meta description one time in three).
• How to identify the sources AI engines actually cite, spot the sources that mention your competitors and not you, and amplify the ambassador URLs that already speak well of your brand.
• The trust and neutrality signals that get content selected: figures, recent testimonials, co-branding and comparisons that name other players. With the Oodrive AI-First™ test, mentioned at the top of the AI answer about 20 days after the campaign.
• The checks to run before publishing anywhere (OAI-SearchBot is sometimes blocked), and how Information Gain and contextual product data earn the citation.

The talk is in French with English subtitles. Fill in the form to watch the full replay.