The starting point: an established trade with no digital footprint
Tuva Euro BV was an established trading operation that ran entirely on personal networks and existing relationships. There was no website, no Google presence and no visibility in AI search tools. In a B2B market where buyers increasingly research suppliers online before making contact, the business was invisible to everyone who searched for what it sells. The full case sets out the brief; this article explains why the sequence worked.
Why a niche B2B supplier can be named quickly
A niche B2B supplier can be named quickly because AI assistants rank businesses they have no memory of on what they retrieve at that moment. Researchers at the University of Toronto tested this with GPT-4o in a study at the EDBT/ICDT 2026 workshops: when they shuffled the retrieved snippets, rankings for well-known brands barely moved (a mean rank deviation of 2.30), while rankings for niche entities shifted almost twice as much (4.15). When they swapped two brand names inside the snippets, the niche rankings followed the text.
For a supplier of poultry meal, that is the whole opportunity. No model has an opinion about the company from training. Whatever it retrieves decides the answer, and what it retrieves is exactly what the business can build.
Step 1: a site built to be retrieved, not just to exist
The first step was building tuvaeuro.eu from scratch with an architecture crawlers could read: structured data naming the organisation, fast pages, a heading hierarchy that names products in the words buyers use, and internal links across product categories. This is the entry ticket on both sides. Google's own guidance says being indexed and eligible in Search is the precondition for its AI features, and an assistant cannot cite a page its crawler was never able to fetch.
Step 2: a keyword map with buyer intent, not volume
The second step mapped the commercial search landscape for poultry meal, liquid eggs and related animal proteins across EU markets, keeping the terms with genuine buyer intent. This is where most B2B sites go wrong: they rank for definitional queries and are never retrieved for "supplier of" queries. The map decided which pages existed and what evidence each carried, so that the pages an assistant retrieves for a supplier question were the pages that named the supplier.
Step 3: the entity, on and off the site
The third step made the business recognisable as an entity: consistent organisation data on the site, and an external presence that AI tools could use to identify and understand the company. This is the signal the research weights highest. In Ahrefs' analysis of 75,000 brands, mentions of the brand across the web correlated more strongly with AI visibility than Domain Rating or branded search volume. For a new domain with no rating at all, that finding is the reason the plan could work.
Step 4: a small number of relevant links
The fourth step was a targeted set of backlinks from relevant industry sources, to establish the early authority signals Google uses in B2B categories. The set was deliberately small. Links drive organic discovery and therefore eligibility for Google's AI features; they are not what gets a niche supplier named by an assistant, so the budget went to relevance rather than volume.
What happened, and what it cost in ads
| Outcome | Result |
|---|---|
| Google position for several commercial keywords | 1 to 3 |
| Time to page-one rankings | Under 6 months |
| SEO and GEO workstreams delivered | 5 |
| Paid ad spend behind the new enquiries | €0 |
| AI assistants naming the business for supplier searches | ChatGPT and Perplexity |
The enquiries came from being findable and being named, not from a campaign. That is the point of the sequence: a business that exists as an entity, with pages built for the queries buyers actually use, is retrieved when the buyer asks, whether the buyer asks Google or an assistant.
What this means for another niche B2B business
The sequence transfers because the mechanism does. A niche business is ranked on retrieved evidence, so the work is to make the right evidence exist and be reachable: a crawlable site, pages mapped to supplier-intent queries, organisation data and external mentions that identify the company, and a few relevant links for discovery. Two cautions belong here. AI answers vary from day to day, and only 34 to 42 percent of cited sources stay the same between consecutive days in the St. Gallen measurement, so "named by ChatGPT" is a rate, not a permanent state. And the timescale was under six months for a business starting from nothing; a business with an existing site and a weak entity usually moves faster. The small business article covers the mechanism in detail.
Frequently asked questions
Did the business have any online presence before?
No. There was no website, no Google presence and no visibility in AI tools. The business ran on personal networks. The site, the keyword map, the entity work and the links were all built within the engagement.
Which AI assistants name it now?
ChatGPT and Perplexity, for supplier searches in its category. Because assistants are probabilistic, this is measured as a rate over repeated queries rather than claimed as a fixed position.
How much was spent on advertising?
Nothing. The new inbound enquiries came without paid ads. That is not a rule for every business; it is what happened here, because the category is niche and the buyers search for suppliers by name of product.
Would this work for a consumer brand?
Less directly. For well-known consumer brands, assistants rely more on what they learned in training, so retrieved content moves the ranking less. The niche mechanism is strongest where the model has no prior opinion.
How long did the GEO part take compared with the SEO part?
They ran together. The site architecture and keyword map serve both; the entity and external presence work is the GEO-specific layer. Separating them would have doubled the editing on the same pages.
Can I see whether my business is named today?
Yes, in minutes. The free analysis queries ChatGPT, Perplexity, Claude and Gemini for your category and market, reads your key pages and reports what the assistants cite instead of you.
Find out where your business stands
The free analysis is the first step this case started with: whether the assistants name you for what you sell, who they name instead, and which of the four steps above your site still needs.
