01 / TECHNICAL BASE
Make the entity crawlable.
I created a minimal site with a clear brand, product, location, robots.txt, sitemap, JSON-LD, FAQ content, and an llms.txt file. The goal was machine-readable clarity, not visual polish.
AI visibility / field experiment
How one citation put a new brand in AI answers.
I tried to make a brand that did not exist visible to AI assistants. The surprising result was not a huge content machine. It was a small, consistent footprint and one credible external mention.
The visibility loop
Consistency creates a usable entity.But visibility is not the same as truth, trust, or commercial success.
The short answer
Can one external citation make a new brand visible to AI?
Yes, it can be enough to trigger a mention in a particular answer. In this experiment, Perplexity surfaced De Schaut Cookies after a Reddit post became the one external source it used. ChatGPT later mentioned the brand after a YouTube video was published.
That is a useful proof of possibility, not proof of durable authority. AI systems are probabilistic, their indexes and retrieval methods differ, and the same prompt can produce different results. The lesson is to build a clear, consistent evidence trail, then test what survives across engines and over time.
Brand: De Schaut CookiesProduct: matcha cookiesOrigin: Ghent, Belgium
The experiment
I started with a brand that had no history.
De Schaut Cookies was a deliberately new concept: matcha cookies from Ghent, Belgium. There was no established reputation, customer base, or editorial footprint to inherit. That made it a useful test of what a small, intentional publishing system could do.
02 / CONSISTENT FACTS
Repeat the same facts.
The wording stayed stable across assets: De Schaut Cookies, matcha cookies, Ghent, Belgium. Clear entity co-occurrence gives a model fewer opportunities to blend the concept with something else.
03 / EXTERNAL PROOF
Seed trusted surfaces.
I added concise references on Reddit, X, and YouTube. Each surface repeated the same core facts while contributing a different kind of discoverable evidence.
What changed
The turning point was not the robots.txt file.
A reachable site proves that a crawler can visit. A credible mention gives the model something to repeat.
Observation from the experimentRobots.txt was a sanity check.
ChatGPT could access the file, which confirmed that the domain was reachable and not accidentally blocked. The brand did not appear in answers just because the file was available.
Reddit supplied the first useful citation.
Perplexity eventually mentioned the brand using one Reddit source. A large footprint was not required for this first signal to appear.
YouTube created another retrieval path.
A short video and a consistent description gave the brand a second external surface. The brand was later mentioned in ChatGPT after that video existed.
What the result does not mean
Visibility is a probability, not a guarantee.
The experiment was intentionally small. It demonstrates that a new entity can enter an AI system's retrieval orbit, but it does not show that the entity will appear for every user, every prompt, or every model.
New brands have an easier test.
There were no established competitors or conflicting signals to overcome. Larger brands need more evidence and more strategic differentiation.
Fresh indexes behave differently.
Perplexity retrieves current web information in real time more visibly than some other systems. Faster discovery does not equal permanent memory.
Incognito testing matters.
Run neutral prompts in a logged-out or incognito session where possible. Personal history, location, and previous conversations can influence answers.
Repeat the measurement.
Test several prompts, engines, dates, and markets. Record citations, wording, contradictions, and whether the answer explains the brand correctly.
This is a field note, not a controlled ranking study. The observations are directional and may change as model indexes, retrieval systems, prompts, and third-party pages change.
Questions buyers actually ask
What should a company do with this?
What is AI visibility?
AI visibility is the likelihood that AI assistants mention, describe, compare, or recommend a company when someone asks a relevant question. It includes whether the answer is accurate, what sources support it, and whether the company appears for the prompts its buyers actually use.
Is AI visibility the same as traditional SEO?
No. Traditional SEO focuses on ranking pages for search queries. AI visibility also depends on retrieval, entity clarity, source quality, answerability, and whether multiple sources support a consistent description. SEO still matters because many AI systems discover and cite web content through search-like processes.
How many mentions does a brand need?
There is no universal number. One credible mention can be enough to trigger an early result for a new brand, as this experiment showed. Durable visibility for a competitive company usually requires a broader, consistent body of evidence across owned content, independent sources, expert commentary, and customer-facing answers.
Should companies create fake brands or fake listings?
No. This experiment used a fictional concept as a test environment, not as a recommendation to fabricate commercial claims. Real companies should publish accurate information, earn legitimate references, disclose affiliations, and avoid manufactured reviews, listings, or testimonials that mislead people or models.
What should I measure first?
Start with a fixed prompt set covering category, use case, competitors, objections, and recommendation questions. Record whether the company appears, how it is described, which sources are cited, what is missing, and whether answers agree across models and dates.
Apply the method
Want to know what AI says about your company?
I can run a focused AI visibility teardown around your category, buyer prompts, competitors, and the evidence shaping the answers.