Week 9 · the seventh dimension

Citation & authority: who AI trusts to describe you

The last dimension is the one least in your hands — the outside sources an AI trusts. It's also where the durable advantage is won.

Makefield · Berlin

The seventh, and the hardest

Whose account of you the AI trusts

The final dimension is the one furthest from your direct control: the external sources an AI draws on when it describes you. And the uncomfortable fact, seen across audits, is that for many companies the cited sources aren't their own site — they're third parties: review platforms, comparison articles, press, and community discussion. The model is describing you through someone else's account of you. Your narrative is, in part, owned by sources you don't write.

What good and bad look like

Current, plural, favourable — or fragile

Healthy citation looks like this: the AI draws on current sources (the last year or two, not three-year-old reviews), from several high-authority places in your category, and the content it's pulling is accurate and favourable. Unhealthy looks like the opposite: citations dominated by a single fragile source, or stale ones, or — the costly case — a comparison article written from a competitor's angle that the model treats as neutral. And remember the split from the field notes: being cited as a source is not the same as being recommended — this dimension is about the sources, the machinery beneath the recommendation.

What the work actually is

Earned, in public, slowly

You can't edit the sources you don't own, so the work here is to earn presence in them — honestly. Be genuinely useful where the model already looks: communities like Reddit carry enormous weight, and the way to be cited there is to actually help, not to plant. Earn current third-party coverage. Correct the record on the review and comparison sites that carry you. This is the slowest part of the method and the least fakeable — which is exactly why it's the durable advantage. Documentation and schema can be matched by anyone in a week; a real, earned presence in the sources AI trusts cannot.

The seven, complete

The whole picture

That closes the seven dimensions: the outcome — recognition, accuracy, competitor framing, recommendation — and the machinery that moves it — documentation, structured data, citation and authority. Together they're a full map of how an AI comes to say what it says about you, and where each problem actually lives. What's left is to measure all seven properly — rigorously, reproducibly, the same way every time — which is where the audit comes in.

Next — The audit: turning these seven dimensions into a measurement you can stand behind. New here? Start with the overview.

Common questions

Questions this lesson answers

Why does AI cite other websites instead of mine when it describes my company?

Because for many companies the sources an engine draws on are not their own site. Review platforms, comparison articles, press coverage and community discussion carry a large share of what a model repeats about you, so part of your account is written by people you do not employ. That is the seventh dimension, and it is the one furthest from your direct control. It does not make your own site irrelevant, since your documentation and schema are still what a model reads when it does reach you. It means the outside record is a separate piece of work with a separate remedy.

How do I get AI to cite my company as a source?

Slowly, and by being useful where the model already looks. Planting posts, buying placements or seeding review sites tends to produce the fragile, one-source pattern that a careful read exposes anyway. What works is earning current third-party coverage, taking part honestly in the communities that carry weight in your category, and keeping the profiles that already describe you accurate. This is the slowest part of the method and the least fakeable, which is why it is also the most durable: documentation and schema can be matched by a competitor in a week, and an earned presence in the sources an engine trusts cannot.

What can I do about a comparison article that describes me from a competitor's angle?

Treat it as a source problem rather than a ranking one. A comparison piece written from a competitor's angle is one of the costlier cases in this dimension, because a model can read it as neutral and repeat its framing as fact. You cannot edit a page you do not own, so the honest options are these: correct anything factually wrong with the publisher or the platform directly, keep your own entry on those sites current and complete, and work on the plural half of the problem so that one article is not carrying your description alone. A single fragile source is the condition that makes one bad article expensive.

If more sources cite me, will AI start recommending me?

Not on its own, and the two are worth keeping apart. Being cited as a source means a model used a page to build its answer. Being recommended means it named you when someone asked who to choose, and audits routinely find that the two lists differ. Citation and authority is machinery: it changes what a model has to work from, which is why it also moves how accurately and how fairly you are described. Recommendation is an outcome that follows from the whole picture, and no single dimension buys it.