How Long Does It Take for a New Brand to Be Cited by AI? A 23-Day Experiment
Summarize this article with AI
23 Days to Exist in the Eyes of AI
23 days. That’s how long it took an unknown entity to be cited by 5 artificial intelligence systems. I read this data on a Friday, between two black coffees at dawn, from my office in Southeast Asia. And I immediately thought of a client—a digital brand that had just been born, with no reputation, and had spent €7,500 on SEO content without a single mention in ChatGPT or Perplexity.
The finding was simple. An eco-responsible ready-to-wear online shop, launched 3 months earlier. 120 product pages, 40 blog posts, few backlinks. On Google, traffic was climbing slowly. But for AI, nothing. No generated response contained the brand name. The founder called me one Tuesday morning. « Stéphane, why don’t we exist for tomorrow’s tools? »
I started with this number: 23 days. Because at the same time, an anonymous researcher had just published an experiment that blew up the usual assumptions about AI indexing. What I read made me reposition my client’s approach in 40 minutes of calling.
The Experimental Protocol That Changes Everything
A researcher created a web entity from scratch. No history. Name, site, mentions: everything started at zero. For 23 days, he queried 5 different AI systems daily with web access: ChatGPT, Perplexity, Claude, Copilot, and Gemini. He asked them for very specific info about this entity. Result: 16,000 data points analyzed. Each response scored +1 if correct, 0 if the entity wasn’t found, −1 in case of hallucination.
The most surprising part? The AI models’ crawlers were blocked. Cloudflare returned a 403 to GPTBot, ClaudeBot, PerplexityBot, and CCBot for 22 out of 23 days. No direct access to the site’s pages. But on day 6, AI was already citing the entity.
| Day | Action | Result |
|---|---|---|
| 0 | Entity creation | No trace |
| 1–5 | Daily queries | Score 0 for all AI |
| 6 | First mention | Correct citation despite block |
| 7–13 | Flow of mentions | Increasingly precise responses |
| 14 | Knowledge Graph appears | AI relies on knowledge graph |
It’s clear: crawled access is not the trigger. The entity appeared via the Knowledge Graph, fed by third-party signals that AI systems aggregated without the site itself.
Crawl Is Not the Trigger. Knowledge Graph Is.
Many e-commerce leaders still think that to appear in ChatGPT or Perplexity responses, their site must be crawlable by training bots. The 23-day experiment breaks that belief. No crawl, no problem. The entity was cited a full week before crawlers ever touched a single page.
The Knowledge Graph is that massive database of entities (people, organizations, products, concepts) that Google and other engines build by aggregating authoritative sources: Wikipedia, Wikidata, government databases, news, patents. Generative AI models consult these graphs in real-time or near-real-time, without needing to read pages on your site.
In the DOSE framework I’m building with Guillaume Attias at BMO Academy, the « E » for Entity is precisely there to structure the brand as an entity recognized by these graphs. When you activate this pillar, you no longer depend on a hypothetical crawl to exist in AI responses. You exist where models draw their authority.
Cette chronologie visualise les étapes clés de l’expérience de 23 jours, montrant quand l’entité est apparue pour la première fois dans les réponses des IA et comment elle a gagné en traction via le Knowledge Graph.
Timeline: 23 Days to AI Citation
From zero to cited by 5 AIs in 23 days
14 Days of Latency, Then the Signal That Flips Everything
The experiment shows a second turning point: between the first correct citation (day 6) and the emergence of a rich representation (day 14), it takes roughly 10 days. During those days, AI systems aggregate indirect signals: word co-occurrences in third-party articles, additions to structured databases, citations via public APIs. They build the entity.
This lag seems short. You’d want a brand name to appear the very evening of launch. But AI builds itself through capillary action, not in an instant. This is what I see on e-commerce launch projects: the first mentions in Perplexity or Google AI Overviews responses typically arrive between day 10 and day 20 after the first structured signal.
The cookie banner and robots.txt file have nothing to do with it. What matters is being recognized as an entity by the Knowledge Graph. And that’s something you build before launching the site.
How a Freshly Launched E-Commerce Brand Broke Through in 19 Days
Back to the client who called me one Tuesday. After the audit call, we stopped producing unfocused content. We concentrated on three workstreams.
- Created a Wikidata page with 14 structured fields, linked to the official site and physical address
- Published a sector press release on an economic news platform, citing the brand and its founder
- Added the site to Google Business Profile with a Knowledge Panel immediately available
19 days later, the brand appeared in Perplexity responses for niche queries. 37 AI queries generated 2,400 additional sessions in 2 months, without an extra penny of paid spend. The €7,500 budget that was going up in smoke became an investment focused on triggering signals.
I’m not selling you the method. I’m showing you the results.
Is Your Brand Already an AI Entity? The 30-Second Test
Before writing an article or touching your sitemap, ask yourself these four questions. If you answer no to any one of them, your entity signals are probably too weak.
1. When I type the brand name into ChatGPT Search or Perplexity, does the response give a definition of the company, not just a link?
2. Is there a well-categorized Wikidata page with a unique identifier?
3. Does the Google Knowledge Panel show up on mobile and desktop?
4. Do third-party sites (press, directories, product databases) cite the brand as a named entity, with a link to the site?
In my audits, these four points are my first test. I still see brands that have invested €10,000, €20,000, €30,000 in « classic » SEO without fixing this. Result: their traffic stalls on Google Search, and AI ignores them.
Conversely, when these four conditions are met, the citation timeline with my clients often drops below 10 days. Even without backlinks.
What You Gain by Structuring Your Brand as an Entity
Goodbye to the dogma of ever-more content. Goodbye to passively waiting for a crawler to pass. Activate three levers from day one: the structured page, the authority mention, and anchoring in the Knowledge Graph.
This approach is the infrastructure that makes a recent brand exist in the AI ecosystem without waiting 6 months. Result: 23 days for a generic entity, 19 days for my e-commerce client, sometimes less when the Knowledge Graph is fed early.
In 2026, SEO is won on entity recognition, not crawl. Every day without your brand defined as an entity in knowledge bases costs you.
The good news: no entry fee, just an architecture to build. And your brand—does it have a birth date yet in the AI index?
Is Your Brand Ready for AI?
A 45-minute live audit. I screen your site for AI entity signals and give you the exact sequence of actions, no theory. First step to turn your brand into a recognized entity.
Book a strategic call — 45 minFrequently Asked Questions
Does my site absolutely have to be crawlable by AI bots?
In the short term, no. I see AI citing an entity via the Knowledge Graph without ever crawling the site. But in the medium term, allowing crawled access gives more precise responses.
What are the triggering signals for a new brand to be cited?
There are three keys: a Wikidata page or Google Knowledge Panel, a mention in a recognized news source, and registration in third-party databases (business directories, product databases).
How long does it take for an unknown brand to appear in AI responses?
I observed over 23 days: the timeline varies from 6 to 20 days after the first signals. The average is around 14 days.
Can you speed up the process of AI citation?
Yes. Create an entity page before public launch: Wikidata, Google Business, and if possible a press article. The earlier these signals appear, the faster AI identifiés the entity.
My site is new. Should I stop classic SEO to focus on AI?
No, both are complementary. Classic SEO brings immediate traffic. Entity recognition positions you for AI engines over the next 24 months.
