Three websites, three companies, one founder named on one of them
By Meet Patel · 2026-08-31 · updated 2026-08-31
Summary
An audit of three related domains found three different descriptions of one company, a product declared as an organisation, a cross-domain reference pointing at an identifier that did not exist, and the founder named on only one of the three.
I went looking for why nothing I write gets cited by any AI system, and found something more basic.
My personal site, my company's site and my product's site each described a different company.
What was actually there
My personal site said the company builds autonomous decision intelligence, named me as founder, and carried a full set of structured data tying the three things together.
The company's own site described itself as building "the next generation of business intelligence" — the category we deliberately moved away from months ago. It had no structured data at all. It did not contain the words "Meet Patel" anywhere. It did not link to my site.
The product's site had structured data, and it declared the product as an organisation. Not a product made by a company. An organisation, named after the product, with no founder and no parent. It also did not mention me.
Three properties I control. Three different companies, as far as anything reading them could tell.
The one that made me wince
My site had been pointing at an identifier on the product's site for weeks. A reference saying "the thing described over there is the product this company makes".
The thing over there did not have an identifier. The reference pointed at nothing.
Nothing broke. No error, no warning, no report anywhere. A pointer into empty space is completely silent, which is why it survived. Everything that fails loudly gets fixed; this class of problem is only ever found by someone going and looking.
Why it matters more than it sounds like it should
My name is common. Search it and you get someone at a large tech company, a doctor in Ohio, a musician, and a page of academic citations. I am not on the first page.
The way out of that is not ranking harder for my own name. It is being resolvable — having enough independent things agree on who this particular person is that a system can tell us apart with confidence.
Agreement is the operative word. My site making the claim three times is one source repeating itself. What resolves an identity is separate properties saying the same thing, and mine were saying three different things.
A system reading all three has one safe move available, which is to hedge. Hedging is how you end up losing to the doctor.
The rule I got wrong
I had assumed each site should describe itself fully. That is the instinct — each property is standalone, so each should be complete.
It is backwards. One property holds the full description; every other one references it by the same identifier and says almost nothing else.
Repeat the description in three places and you have three copies that will drift, because they are maintained on different days by different people with different priorities. Mine drifted by a whole category. Reference it in three places and there is only ever one version to keep true.
Check yours in about ten minutes
Open your company site, your product site and your own site side by side.
- Do all three describe the company the same way? Same category, same words.
- Is the company's name spelled identically on all three, including the space?
- Is the founder named on more than one of them?
- Does each link to the others in visible copy, not only in markup?
- If any of them carries structured data, does every reference in it point at something that actually exists?
I failed four of those five. The one I passed was that all three were online.
The part markup cannot fix
The fix I reached for first was structured data, because it is tidy and I can write it in an afternoon.
The more useful fix is one sentence of visible prose on the company site: founded in Dubai by a named person, with a link. Machines read markup. They also read the page, and they weight the page more, because prose is harder to fake at scale than a script tag.
The markup took twenty minutes. The sentence took two, and I suspect it will do more.
Frequently asked questions
Why do AI models get my company wrong?
Usually because nothing outside your own site corroborates what your site says. Models resolve entities from agreement across independent sources, so when a company's own domain, its product domain and its founder's site each describe things differently, the safest thing a model can do is hedge.
Should each of my sites repeat the full company description in structured data?
No. One property should hold the full description and every other should reference it by the same identifier. Repeating the description in three places creates three versions that drift apart, which is the contradiction you were trying to avoid.
Written by Meet Patel — founder of Company 8, building Dan (usedan.com). Dubai, UAE.