89% of AI Search demand has no owner: how your e-commerce can seize it before the window closes
Summarize this article with AI
Cette répartition illustre le constat clé de l’étude de Kevin Indig : la quasi-totalité des requêtes générées par ChatGPT n’a pas encore de marque de référence. Le vide est immense, et la fenêtre pour le combler est ouverte.
89,3 % de la demande IA sans propriétaire clair
Une opportunité massive pour les e-commerçants qui agissent maintenant
1,094 catégories, 600,000 citations, a truth that stings
I’m reviewing Kevin Indig’s data this morning. His analysis spans six months of ChatGPT responses, compiled via Semrush. January through June 2026.
The dataset is substantial: 1,094 U.S. catégories, more than 50,000 brands, 600,000 citations.
And the finding is clear.
89.3% of AI search demand has no clear owner.
Think about it. Out of 100 informational or transactional queries typed into ChatGPT, 89 don’t point to a dominant brand.
No leader.
No established brand.
A field of ruins—or rather, an opportunity.
Indig built three groups:
- Clear owner: a brand cited in at least 4 prompts out of 5, with a 5-point lead over second place.
- Emerging leader: 3 prompts out of 5, but without the decisive gap.
- Open territory: no brand leads across 3 prompts.
Result: only 15.2% of catégories have an owner.
53.7% are completely open fields, with several credible brands but none in command.
More striking: the catégories with high AI search volume—those concentrating 98% of demand—show an even lower ownership rate: 11.3%.
The money is on the table, and nobody is picking it up.
Kevin Indig says it himself: « Topical authority matters in AI search. Once a brand gets an outsized share of voice, it keeps it. » But right now, keeping isn’t the problem. The challenge is conquest.
The window is wide open. For how long?
The AI Search market is still young. Not all brands have grasped that citations count. Or that a simple architecture shift can flip a category.
A client calls me in March 2026. He runs an auto parts e-commerce. 8,000 SKUs, stable Google organic traffic, but zero mentions in ChatGPT responses on his key catégories.
Yet his site has content. Category pages, product sheets, maintenance guides.
The problem? No semantic structure.
No semantic cluster.
No clear signal for models.
We applied a simple overhaul: entity grouping, dense internal linking, enriched pillar pages with comparisons and structured data. In 4 months, the site began appearing in ChatGPT answers for queries like « best diesel SUV oil filter brand. »
In 8 months, citations climbed 340%. Overall organic traffic gained 22 points. But most importantly, the AI Search channel became a source of qualified leads, with no marginal cost.
Why does it work?
Because language models seek thematic coherence signals. A site that covers a topic in depth, with pages linked to each other and well-defined entities, becomes a natural reference. No ad budget needed. Just a structure that speaks the language of algorithms.
Almost nobody polishes this structure. Result: 89.3% of the pie is still up for grabs.
Topical authority: the only durable shield
Indig puts it well: once a brand locks in a category, it keeps it. Over 6 months, the numbers are clear: gaps widen, slowly but surely. The early bird isn’t always the best armed, but it has a definite advantage.
In June 2026, in the rare catégories where a clear leader emerges, that leader is cited nearly 70% of the time. Second place caps out at 15–20%. The gap is huge.
This result comes from the citation mechanism: each mention reinforces the signal. The model returns to it mechanically. It’s a virtuous circle.
I observed this with another client, a design furniture e-commerce. In 2025, they started structuring a cluster around « Scandinavian convertible sofa. » Back then, no player dominated this category in AI. 14 months later, they’re cited in 4 prompts out of 5, with a 22-point lead over second. Their competitors, despite ranking better on traditional Google, stay invisible in AI.
Why?
Because they scattered their efforts. Too many catégories covered at surface level, no semantic backbone.
The algorithm picked the most coherent source, not the biggest.
Voici le processus que j’applique depuis des années pour les clusters Google, et qui fonctionne tout aussi bien pour l’IA. Chaque étape renforce la suivante pour construire une autorité thématique durable.
Les 5 étapes pour bâtir un cluster sémantique qui parle à l’IA
De la sélection de catégorie au suivi des citations, une méthode éprouvée
Build a cluster that speaks to AI (and Google too)
The method I’ve applied since 2016 for semantic clusters is the same one that works for AI Search. At heart, the principle is identical: show topical authority through structure. Here are the steps.
1. Pick a high-volume category with no AI leader.
I use Semrush or Ahrefs to spot product catégories generating volume in AI. I cross-reference with Indig’s data: I target niches where ownership rate is below 15%.
2. Map the entities.
I list all key concepts: products, brands, use cases, problems, comparatives, technical attributes. These entities become my cluster pillars.
3. Create pillar pages.
One pillar page per category, covering the entire topic with links to child pages. Each child page digs into a subtopic in depth. The whole thing looks like a knowledge network, not a product sheet list.
4. Precise internal linking.
Each child page links back to the pillar and to relevant sibling pages. I use descriptive, varied anchor text. AI loves contextual links.
5. Structured data and markup.
I implement Product, FAQ, HowTo, BreadcrumbList schemas. I give machines clear hints about content type.
6. Measure citations.
I query ChatGPT regularly with my target prompts. I note my brand’s position versus competitors. I track the shift over 3, 6, 12 months.
These steps don’t need a colossal budget. They need clear vision and consistency. I’ve seen 50-pillar sites crush 5,000-page sites purely because their architecture was coherent.
378% more citations, without touching the ad budget
Back to the auto parts client. Before we stepped in, he had 12 category pages. All indexed, but isolated. No pillar pages. Maintenance guides were scattered across a blog with no links to products. The site talked about « air filters, » « spark plugs, » « belts, » without ever tying them together.
We rebuilt it. We created a pillar page « Engine Maintenance for Diesel SUVs » grouping 18 child pages. Each child detailed one part, with comparison tables, price ranges, and cross-links.
Result in 8 months:
- ChatGPT citations: +378% (from 4 monthly citations to over 19).
- Overall organic traffic: +24%, with a 47% boost on SUV-related pages.
- Revenue from AI Search channel: 31% of the period’s incremental revenue.
Most interesting: his main competitor, a generalist marketplace, never succeeded in AI despite a budget ten times higher. Why? Because the marketplace mixed genres: auto parts, home improvement, gardening. Impossible for AI to spot a precise theme.
This case isn’t isolated. I see it every week in audits: thematic specialization pays. Not site size. Not brand fame. Coherence.
Act before the market understands the mechanics
Kevin Indig’s study dropped July 20, 2026. It’s still fresh. Very few e-commerce players have woven it into strategy. Most are waiting for « AI Search to become a real channel. » That’s why you have an edge.
Waiting means letting competitors grab the vacant catégories. And once a leader emerges, the gap widens. Indig proves it: positions stabilize after a few months. The window isn’t forever.
I’m not selling you the method. I’m showing you the pages.
The AI Search audit I offer starts with a simple fact: which 5 catégories could your site own as AI leader in under 6 months?
We identify catégories with no owner.
We map missing entities.
We sketch the cluster skeleton.
And we deploy.
The cost is trivial next to qualified traffic potential. And ROI measures in weeks, not years.
You can keep betting on Google alone. But Google itself integrates more and more AI answers. Classic organic traffic is shrinking. Demand shifts to assistants. 89.3% of that demand is yours to take. Now.
One question should guide your next moves
When I talk to an e-commerce leader, I always ask the same thing: « Does ChatGPT cite your brand as a reference on your top three revenue catégories? »
The answer is almost always the same. Silence. An awkward smile. « I don’t know. »
That silence has a cost. 89.3% of qualified audience doesn’t even know your brand exists.
AI Search isn’t a gadget. It’s the new influence channel. Ignoring it today is like ignoring mobile in 2010. By 2012, some got it. By 2015, it was too late.
So I ask again: does ChatGPT cite your brand as a reference on your flagship category?
If you can’t answer « yes » with numbers, you’re probably part of the 89.3% left behind.
What will you do today to break out?
AI Search audit: let’s identify your catégories to conquer
I’ll deliver, in 45 minutes, the 5 AI catégories still without an owner where your e-commerce can dominate. Live audit, concrete action plan, ROI estimate. This isn’t a pitch—it’s an operational roadmap.
Book a strategic call — 45 minFrequently Asked Questions
How do I know if my category already has a clear owner in AI Search?
Run 5 different prompts on ChatGPT for your category. If the same brand appears in 4 prompts out of 5 with a clear gap, it’s the leader. Otherwise, the field is open. Use Semrush to automate tracking.
My site is small. Can I really compete with big brands on AI Search?
Yes, because AI doesn’t measure site size or brand fame. It looks for thematic coherence. A small, highly specialized site with a solid semantic cluster regularly beats generalists. I’ve seen it with my auto parts client.
Do I need to create AI-specific content, different from Google content?
Not really. The same solid cluster works for classic SEO and AI Search. What counts is structure: clear entities, precise internal links, structured data. Invest in architecture, not duplicate content.
How long does it take to become the leader of a category in AI?
I see first results after 3–4 months, dominant position around 8 months. It depends on competition and linking quality. The earlier you start, the wider the gap with second place grows.
Is topical authority enough, or do I also need backlinks for AI?
Indig’s study confirms it: AI citations aren’t tied to backlinks or Google traffic. Topical authority and semantic coherence are what truly matter. Links still help overall SEO, but AI looks at structure first.
