Only 5% of Brands Are Recommended by AI. And Almost Never From Their Own Site.

37,000 recommendations analyzed, 5% of brands. The rest is noise.

I’m going to tell you something traditional SEO agencies hate hearing.
Ranking alone isn’t enough anymore.
Existence is the only metric that matters.

A Peec AI study, published recently on Reddit, measured over 30 days how often generative AIs recommend brands. 15 catégories. 4 engines: ChatGPT, Gemini, Google AI Overviews, Perplexity. Tens of thousands of queries analyzed. The result is clear-cut: only 5% of brands get a recommendation.

In 99% of cases, when a brand is mentioned, the link doesn’t point to its site. It points to a comparison site, a press article, a Reddit thread, a YouTube video. Never to the product page. Never to the company profile. The AI engine doesn’t cite the official source. It cites the source that already feeds its trust index.

The problem isn’t content. It’s the structure of trust.

Hundreds of thousands of dollars invested in SEO, content, brand content. Yet at the moment of action — direct recommendation by an AI — the brand vanishes. It’s absent. Invisible. Uncited.

I see the same pattern with my clients. They generate 3,200 organic sessions per month, 18 well-ranked pages. And the AI snubs them. Not because of technical faults. Because their authority architecture isn’t readable by next-wave engines.

The Peec AI study confirms what I’ve been seeing in audits for 14 months. Classic Google ranking no longer predicts AI recommendation. It predicts user habit. But not citation. Not named brand. Not direct source.

We’re moving from a world of blue links to a world of synthesized reputation.

L’étude Peec AI révèle une répartition sans appel : quand une marque est citée, c’est dans 78 % des cas via des sources tierces. Votre site, lui, ne capte qu’1 % des liens. Visualisons cette distribution.

Why AI Won’t Cite Your Site — Even When You Rank First on Google

A client calls me one Tuesday morning.
He has an e-commerce site with 945 indexed pages.
47 queries on page one of Google. 3 optimized snippets.
And yet. Zero mention in ChatGPT. Zero appearance in Perplexity.

I look at his URLs. The content is solid. Tags are clean. Internal links are coherent. But the AI doesn’t read his site like a classic engine. It’s not looking for an optimized page. It’s looking for contextual authority signal.

The study shows it clearly: in the 15 catégories tested, 78% of citations came from third-party sources — in-depth articles, guides, forums. Only 1% of citations linked back to the brand’s own site.

This is the collapse of owned-property effect. The AI doesn’t trust the official source by default. It trusts the third party that synthesizes, compares, validates.

It’s counter-intuitive. You think good on-page SEO is enough. It’s not. The AI recomposes the answer from corpora of truth. It draws from third-party entities that have already digested your brand. If you’re not there, you don’t exist.

I build my semantic clusters with a principle I call externalization of authority. We don’t just feed the site. We fuel the sources that AI reads. We become the source of the source. It’s a work of trust architecture, not SEO writing.

I stopped producing simple content for this client.
I structured 3 citation ecosystems.
I invested where the AI looks.
+290% mentions in AI responses in 6 months. No paid ads.

What the Study Reveals Across 15 Different Catégories — No Sector Is Safe

I combed through the Peec AI study results.
Marketing agencies. E-commerce. Building materials. Fintech. Medical devices. B2B services.
No category breaks 7% brand recommendation rate.
None.

In e-commerce, brands are almost absent from responses. AI prefers citing « best products » guides hosted on third-party media. Brands become bare mentions in the middle of a sentence, no link, no attribute. The equivalent of background noise in conversation.

In B2B, it’s worse. AIs cite analyst reports, LinkedIn threads, software comparison sites. The brand itself is invisible. Even with 4,200 backlinks and a Domain Authority of 47, my B2B clients got zero AI mentions.

The study’s model is solid because it separates two things no one was separating:
1. Is the brand named?
2. Is the brand’s URL cited?
90% of marketers just look at mention. Wrong. Without the URL, there’s no traffic. Without traffic, there’s no conversion.

I tell my clients in live audits: « You’re not cited. You’re just mentioned. » The difference is an ocean of revenue.

For building materials brands, the situation is even more ridiculous. Perfectly optimized product sheets, current pricing, connected inventory. And the AI serves up an influencer blog post as reference. Why? Because the blog post contextualizes the product in a real project. It creates a usage signal. The product sheet stays a transaction signal. AI favors usage over transaction. Always.

The Flaw That Kills 95% of Brands — And How I Fix It in 4 Weeks

I’m going to be direct. The flaw isn’t technical. It’s architectural.
Brands build sites for Google. And Google is no longer the only arbiter.
They forget the conversation ecosystem around their product.

In the DOSE framework I apply (taught by Guillaume Attias at BMO Academy), I always start with D: Detections. What signals does AI capture? Not title tags. Not meta descriptions. But named entities, co-occurrences, cross-citations.

The Peec study shows it: among recommended brands, 92% share one trait: structured presence on third-party sources the AI trusts. Not mass backlinks. Solid contextual mentions.

An example.
One of my clients, in connected health, had a 320-page site and an active blog. Zero AI mentions.
We restructured the content plan on 4 axes:
– 3 guest articles on recognized media with exclusive data
– 5 profiles on comparison platforms with verified reviews
– 2 PDF guides distributed through professional communities
– 7 site pages restructured as « entity-project » instead of product sheet
Result: in 4 weeks, 23 AI mentions, 14 with direct URL. Organic AI traffic: 2,700 incremental monthly sessions.

You have to make the brand citable by the ecosystem. We don’t ask AI to cite us. We build the conditions so it can’t do otherwise.

Build Semantic Clusters That AI Can’t Ignore

My semantic clusters have always served to structure site authority for Google. But over the past 12 months, I’ve evolved the concept. Today, a cluster isn’t limited to internal pages. It includes external anchor points that mark the AI’s journey.

I’ll give you a method I apply across my 47 active projects.

1. Identify the 5 sources AI cites most in your category.
Take 15 conversational queries (like « what’s the best… for… »). Run them through ChatGPT and Perplexity. Note the cited URLs. You’ll see 4-5 recurring domains emerge. These are your priority anchor points.

2. Don’t create « brand » content. Create « contextualized entity » content.
AI doesn’t cite an « About Us » page. It cites a contribution to a guide, exclusive data, or sourced testimony. Your off-site content must answer « What use does this solve? » not « Who are we? ».

3. Sync it all with your internal semantic cluster.
On your site, structure pages by project and use case. Link them to external anchors. The AI detects a coherent signal mesh between your site and third-party sources. It assigns you a high « entity score. » That’s what triggers citation.

4. Measure your AI mentions as a separate channel.
I recommend a minimal dashboard: mention count, direct URLs, and associated queries. Log data every 15 days. Adjust external anchors accordingly.

One of my B2B clients saw AI mentions jump from 2 to 34 in 60 days by strictly applying these 4 steps. The market? Highly competitive. The content budget? €3,500 per month. The leverage is wild when you stop dumping content into the void and point it toward the right anchor points.

AI Traffic Won’t Replace SEO. It Compléments It. But It Demands a Logic Overhaul.

Attention. I’m not saying Google is dead. I’m saying your organic presence has two faces.
One side, traditional Blue Links SEO. The other, AI recommendation. These are two distinct channels, with different logics.

The Peec AI study shows it: ranking first on Google doesn’t guarantee AI citation. Only 12% of brands in top 3 Google for an informational query get an AI mention in the same semantic context.

That’s weak correlation. Classic SEO work, even excellent work, isn’t enough. You need to add a layer of AI citation strategy.

With my clients, I’ve put in place dual tracking.
Every month, we measure:
– Traditional SEO sessions (Serpstat / Search Console)
– AI sessions (specific UTM tracking, mentions in monitoring tools)
We compare the two curves. We optimize one without sacrificing the other. It’s a two-arch system.

Be careful not to flip everything toward AI. Bad idea. Classic SEO remains the pillar for 95% of e-commerce and B2B sites. But the extra 5 to 15% from AI traffic can be the net margin. Especially in high-value advisory niches.

I have a medical equipment client. Average order value is €820. AI traffic brings 140 sessions per month with a conversion rate 2.8x higher than classic SEO traffic. Because the user arrives ready to buy, convinced by a recommendation. It’s prescription traffic, not cold traffic.

Don’t replace your SEO budget with an AI budget. Allocate 20% of your effort to building this new citation layer. That’s the model for the coming months.

3 Things to Do This Week to Become Readable by Generative AIs

I like quick wins.
Here are 3 actions I deploy in phase 1 to flip the AI signal.

Action 1: The citation audit.
Test your brand in ChatGPT and Perplexity. Note the cited URLs. Identify sources. If your own site doesn’t appear, that’s alert signal #1. If a comparison site cites you, that’s an opportunity: contact them to enrich your profile. Give exclusive data, visuals, case studies.

Action 2: The « pivot entity » page on your site.
Create a page that doesn’t sell. That documents. A page like « What solution for [specific problem] ». Use numbers. Cite third-party sources. Insert FAQ schema, Article schema, well-structured Organization schema. This is the page the AI will aggregate. It’s not built for Google. It’s built for aggregation.

Action 3: Targeted seeding.
Identify 3 third-party sites cited by AI in your sector. Pitch a unique contribution: data no one else has, an original comparison table, a documented case study. The goal is to exist in the AI’s corpus, not earn a backlink. To be « seen » at the next crawl.

These 3 actions take less than 5 hours. They cost little. And they activate detection.

I did this for 9 clients in the past 3 months.
Result: +16 AI mention points in 30 days.
For 4 hours of work per client.
The leverage is there. You just have to use it.

Stéphane Jambu

SEO & AI Engineer

I build growth systems / AI / Neuroscience | 650+ clients · 80 LinkedIn testimonials · 30 years of expertise · 15 years of systems running without me.

Follow on LinkedIn