How identity, influence and intent decide what AI says about a business
Written by
Odin Lund
Category
AI search
Date
Share this article
Identity is what the sources agree on
A business does not have one identity in AI search. It has as many as its sources imply. The model assembles an answer from what it retrieved, and where those sources agree it repeats the claim with confidence. Where they conflict, it hedges or drops the claim.
We traced 4,222 answers in one category back to the sources behind them. Three factors separated the businesses that were named from the ones that were not, and the strongest was consistency: businesses whose facts matched across their site, their listings and their reviews were repeated, while businesses with conflicting facts were hedged.
Identity here is not a visual system or a tone of voice. It is the set of checkable facts — what the business does, for whom, where, since when, to what standard — stated the same way everywhere a model can read them.
Influence is upstream of the answer
You cannot edit an AI answer. You can edit what it is assembled from.
That makes influence a question of sources rather than of persuasion. The work is to correct the facts on the pages and listings the business controls, publish answers to the questions customers actually ask in the formats the models retrieve, and earn presence in the third-party sources the models already cite in that category.
In most categories a small number of domains account for most citations, so the useful question is narrow: which sources do the models rely on here, and is the business present on each with the same facts?
Intent changes the answer
The same business is described differently depending on what the customer asked.
In our legal data set, decision-stage prompts — where the customer is ready to choose — named the client in 37.8% of answers, while comparison prompts named it in 6.3%. The shortlist also forms early: the businesses named in the first-query answers were, with few exceptions, the businesses named later in the journey, and businesses absent from the first answer rarely re-entered.
That is why we read every measure by stage as well as by model. A business can hold a strong share of voice at the decision stage and still be missing from the question that forms the shortlist in the first place.
What this means in practice
Start from the questions customers actually ask, by stage, and run them on each model. Record whether the business is named, what is said, and the source behind each claim.
Then work in the order the answer is built: fix the facts so the sources agree, publish the answers those questions require, and earn the third-party sources the models cite. Re-check daily, because the answers move as sources move and competitors publish.
Our team decides what to change and why. Our agents do the reading, the re-checks and the measurement that keep it current. The discipline is the same by hand, only slower.
Share Blog
