If you monitor your business in AI answers, you need observations tied to the names and questions you actually track. AI Brand Mentions applies your campaign's owned-brand labels to the prompt answers already collected. It shows you where a tracked question names your business, and how that moves across collection dates.
Enter the names before collection, not after
The Organizations tab holds the brand names and hosts used to label results. Put in the names buyers actually encounter: product names, former names, common abbreviations.
Most reports apply their labels when the report is drawn, so an organization added today changes how your existing history reads. No new collection required.
Google AI Mode is the exception. Its mention extraction runs off the tracked names, so an organization you never entered cannot show up there as a detected brand. Enter your names before the collection runs, not after you spot the gap.
Prompts are collected once for the whole account
Tracked prompts are account objects. They get collected once, then interpreted using the organizations in whichever campaigns they are assigned to.
Assigning a prompt to a second campaign therefore costs no extra collection. And two campaigns can label the same stored answer differently. That is correct behavior when an agency describes one market for two clients.
An appearance is not a recommendation
The history records position. The answer holds the context. The report will not merge those for you and should not try.
A business named as too expensive for the job still appears in the answer. So does one named as a poor fit for the buyer described. So does one mentioned only as the competitor to whatever got recommended.
Read the wording before you call an appearance a recommendation. Counting mentions without reading them measures presence and not reception, and a client who opens one of those answers will take the report apart.
Compare competitors under the same questions
Use AI Competitor Mentions to look at competing organizations under the same prompts. Keep the provider and the date attached to every comparison you make.
This is the most reliable reading the product gives you, because both brands sat through the same conditions in the same runs. A competitor named consistently where you are named rarely is a real finding, even though every individual answer moves around.
Read a series, not a run
Assistants answer the same question differently on different days. One collection is one observation of a system that varies.
What holds steady across those answers is consistent presence, consistent absence, and direction across weeks. Absent in one run is not a problem to solve. Absent across a month is.
During your brand review, use the dated answer records to identify a description or change that needs closer inspection.
See AI rank tracking, why repeated answers vary, or how citations differ from mentions.