Guide
How AI mentions affect brand visibility
An AI mention is when ChatGPT, Gemini, Perplexity, Grok, or AI Overviews name a brand without being fed that brand in the prompt. Mentions affect brand visibility because buyers who ask assistants never see your rank. Agencies should count unbranded mentions, name who won the rest, and send a brief — not a clip farm.
Northline Agency
Weekly citation brief
Northstar · Week of 8 Sep 2026
12/20
Named in 12 of 20 buyer questions this week.
Recommended in 7 of 20.
best project management software for agencies 2026
Named · Northstar
Asana alternatives for agencies 2026
Missing · ClickUp
Northstar vs ClickUp for client work
Named · Northstar
Monday.com vs Northstar which is better for client retainers
Missing · Monday.com
Next: Write a comparison page for Asana vs Northstar
Prepared by Northline Agency · 12 Sep 2026
What an AI mention is
Count a mention when the model names the brand on a question a buyer would ask. “Best CRM for mid-market manufacturers” that lists the client counts. “Does ChatGPT mention {Client}?” does not count in the headline score.
The ChatGPT mention report for clients is the one-page table. This guide is why the table matters.
Why mentions matter
Mentions are not traffic until someone clicks a citation. Report mentions as mentions. Promise clicks only when you can measure them.
- Shortlist: if the assistant never names you, you are not in the consideration set for that question.
- Trust: a third-party phrasing (“people often use X”) is closer to PR than to a paid ad.
- Displacement: a mention of a rival on your category question is lost demand you will not see in GSC.
Mentions vs rankings
A brand can rank and still be unnamed in ChatGPT. The reverse also happens. That disagreement is the point of a four-engine brief. See AI search visibility.
| Classic ranking | AI mention | |
|---|---|---|
| Unit | Query + URL + position | Question + engine + date |
| Win | #1–3 on the SERP | Named or recommended in the answer |
| Proof | GSC, rank tracker | Dated answer sample |
| Client artifact | SEO dashboard | Weekly report, white-label with the Agency add-on |
How marketers monitor mentions
Doing that by hand is a half-day per client. CiteBrief generates the same letter. Methodology stays visible so a CMO can challenge a row.
- Write 15–30 unbranded buyer questions with competitors in the set.
- Run them on ChatGPT, Gemini, Perplexity, Grok, and AI Overviews.
- Mark named, recommended, who won, cited URL.
- Exclude branded checks from X of N.
- Send five actions. Stop.
Acting on gaps
A gap is a high-intent question where a rival was named and you were not. The action is almost never “prompt harder.” It is a comparison page, a listing, a roundup, or a sales one-pager the model might cite later. Recheck next period on the same prompt.
What not to do
- Add mention counts to Search Console
- Treat one ChatGPT run as the quarter
- Demand #1 on every prompt
- Ask for daily screenshots
FAQ
Is a citation the same as a mention?
No. Mentioned = name in the answer. Cited = a URL used as a source. Recommended = the model tells the buyer to use them. Record all three.
How many mentions is good?
There is no industry index you should pretend exists. Compare to last period and to named competitors on the same questions.
Do PR teams care about this?
Yes. An assistant naming a rival is a third-party mention you used to only catch in press. AI reporting for PR agencies.
Where is the template?
The AI search report template is the page list. The sample report is a filled letter.
Build the mention table for one client this week, or generate it from a domain in CiteBrief.
Free to start · Starter $39/mo · weekly checks · Agency add-on $49/mo
Start a 14-day trial. One client, one brief.