
Authority is not a domain score. It is being named where the models already read.
Covering Trade press, Review platforms, Roundup listicles and Wikipedia.
Link building for AI visibility is the practice of earning placements on the specific domains engines already draw on when answering your category's questions.
It differs from traditional link building in what it optimises for: a mention on a source a model cites is worth more than a followed link on a domain with a high authority score that no model has ever quoted. The target list is derived from engine citations rather than from a domain rating filter.
Earned coverage in trade and technology press, pitched on something genuinely newsworthy. These are the domains models reach for first when they need an independent source.
G2, Capterra, TrustRadius and the ones specific to your category. Heavily cited when a buyer asks which vendor to choose, and structured in a way models parse easily.
The "best tools for X" roundups that dominate commercial queries and get quoted almost verbatim by assistants. Absence from these is a visible gap in any answer.
Bylined pieces from your team on publications your buyers read, which build the author entity alongside the brand one.
Where notability requirements are genuinely met. It is a foundational source for entity resolution, and a page that fails review does more harm than no page at all.
Places that already discuss you without linking or naming you consistently. The cheapest authority available, because the relationship already exists.
We take the domains the engines actually quote in your category and build the list from those, not from a domain rating threshold.
Some cited domains will never cover you. Saying so at the outset is more useful than a target list that quietly never converts.
Pitched on merit, with something worth publishing. No networks, no reciprocal schemes, no paid placements dressed as editorial.
The measure is whether the engines start drawing on the placement, which is a different question from whether it passed authority.
| Point of comparison | Placement on cited domains | Volume link buying |
|---|---|---|
| How targets are chosen | Domains the engines already quote in your category. | Anything above a domain rating threshold. |
| What is being bought | Nothing. Placements are earned and pitched on merit. | Links, usually through a network or a marketplace. |
| What success means | An engine starts citing the source that now names you. | The link count and the average domain rating rise. |
| Volume | Low. A handful of correct placements per quarter. | High, and mostly on domains no model has quoted. |
| Risk | The placement does not land and you have spent the pitch. | Manual action, and an authority profile that has to be disavowed later. |
A placement counts when an engine starts drawing on it. Links are tracked, but the reported measure is citation, against the same fixed prompt set as everything else here.
The instrument is documented on the GEO page.
Of the domains the engines quote for your category, the proportion that now name you.
How many earned placements go on to be drawn on by an engine, which is the number that tells you whether the target list is right.
It is the wrong measure and we do not sell against it. A placement on a domain the engines quote in your category is worth more than twenty on domains they have never drawn on. We report against cited source coverage instead.
A fixed prompt set, five engines, and the competitor set declared before anything is measured. You keep the roadmap either way.