.png)
When an AI answers a question like “What are the pros and cons of the iPhone 15?”, it doesn’t just consult Apple’s official website. It looks elsewhere:
- Forums (Reddit, Stack Overflow…)
- Customer reviews (Amazon, Google Shopping, Trustpilot…)
- Comparisons published by tech media or on YouTube
Why?
Because these types of content are seen as more authentic, more nuanced, and closer to real user experience.
And most importantly, because they’re abundantly available across websites perceived as authoritative and trustworthy.
That’s exactly what language models seek to produce credible, synthetic answers.
As Thibault Renouf puts it:
“AIs will respond based on what they most frequently read on the internet. If you flood the web with biased content, you influence their answers. And they know it.
Historically, companies have invested in owned content: websites, blogs, white papers, and content campaigns.
But those efforts face three key limitations:
- That content is often seen as promotional (and therefore less trustworthy for AIs),
- It’s sometimes poorly indexed or structured (so harder for AI agents to read),
- It’s rarely cited on third-party platforms (and thus less present in AI training corpora).
In contrast, AIs increasingly prioritize external signals such as:
- Google, TripAdvisor, Amazon reviews, consumer forums
- Reddit discussions, TikTok or YouTube comments
- Spontaneous testimonials on social media
These are interpreted as credibility markers.
As Thomas Spitz highlights in the podcast:
“What we’re seeing is a shift from SEO to REO: Reputation Engine Optimization. What matters isn’t just what you publish, but what others say about you.”
.png)
-min.png)
