Brand vs. Non-Brand Reporting in the Age of AI

Brand vs. non-brand reporting has become the most important split in marketing numbers. AI Overviews now appear on most brand-name searches in the US, AI visibility scores get padded by branded prompts, and CAC is flattered by customers who would have bought anyway. One rule covers all three: report the number that describes people who don't know you yet.

Key takeaways

• Branded prompts, branded searches and repeat customers all produce pleasant numbers that say nothing about growth.
• AI Overviews now answer most brand queries, so brand SEO is no longer an automatic safe bet.
• Track AI visibility on a stable set of at least 30 prompts and always show the non-branded result separately.
• Judge acquisition with holdout tests, not blended CAC.
• Agencies need AEO as its own service with measurable reporting.

Three dashboards, one trap

They look unrelated. An AdWeek panel (Shopify, Blue Bottle Coffee, Naked & Thriving, Petzey) agreed that CAC is harder to calculate than ever because the customer journey is fragmented. Semrush's dashboard guide shows how an AI visibility score flatters you once prompts that include your brand name are counted. And the AI Overview data shows that even brand searches are turning into an answer you don't control.

The mechanism is identical each time. You measure people who already know you and present it as proof that new people are finding you. Splitting brand from non-brand, or new from repeat from incremental customers, is a defence against your own optimistic spreadsheet.

AI Overviews on brand searches: what the data says

According to Ahrefs (via Search Engine Journal), the share of brand-name queries triggering an AIO for 100 global brands from Interbrand's Best Global Brands 2025 jumped from 5 to 63 out of 100 between July and September 2026. In a broader sample of 92.5 million brand keywords, the share was 73.3% in September versus 61.3% in July. Later readings sit around 80% and higher.

What that means in practice:

• AIO most often sits at position 1 (48 of 63 cases), taking the spot your own site used to hold.
• It cites Wikipedia most (71.9%), then YouTube (38.6%) and LinkedIn (22.8%). Channels brands often neglect now shape what gets said about them.
• Only 3.4% of sampled answers mentioned a competitor. The bigger risk is outdated or inaccurate claims, not rivals.

Measurement has gaps too. The branded-query filter in Google Search Console (available since March 2026) doesn't allow a custom definition of "brand", and the AI Assistant channel in GA4 doesn't separate AIO visits from ordinary organic traffic. Until that improves, you have to fill in the picture yourself.

How to build a brand vs. non-brand AI visibility report

Semrush describes a process that takes about 90 minutes for the first build and 20 minutes a week to maintain. The core works with any tool.

1. A stable prompt set

Take at least 30 prompts in 5 buyer-journey groups: customer problems, discovery, comparisons, product selection and branded. That gives roughly 200 observations a week. Keep the set unchanged for at least a month, or the trend line means nothing.

2. Non-branded on its own

Branded prompts will always show your brand. The non-branded number answers whether AI recommends you to people who aren't searching for you by name. Keep the branded group only as a sanity check that AI knows you exist.

3. Beware small groups

With 6 prompts per group, changing a single prompt moves the result by 17 points. For stable group-level reporting, scale to 15+ prompts per group.

4. Pair citations with clicks

Semrush gives an example of a page cited by AI across 30 prompts that had zero Google clicks. That is a clear SEO and AEO opportunity. Add your own classification of brand queries and dated AIO screenshots for the client's brand name.

The same logic in acquisition: new, repeat, incremental

Naked & Thriving excludes past customers from acquisition campaigns on Meta, Google and ChatGPT, and drives repeat purchases through email and SMS, which cost less. The key distinction is between a new customer, a repeat customer and one you would not have won without the campaign. Holdout tests are the best way to measure incrementality.

Shopify offers Shop Campaigns (pay per sale, with separate rates for new and existing customers) and Autopilot for cross-channel testing with automatic budget allocation. The panel's advice is to reserve a test budget before peak season so results can be applied when impact is highest. We've covered incrementality on Meta before. The general rule here is simpler: clean up who you are measuring first.

What agencies should do about it

HubSpot notes that clients are raising AI visibility themselves, because they see competitors in ChatGPT or Perplexity answers. "We do that within SEO" no longer holds up. A practical minimum:

• AEO as a separate line item with its own report: positions in AI answers, share of voice against competitors, cited sources.
• Content remapped around buyer questions rather than just keywords, with short definitions, data points, tables and schema. Pages that win featured snippets have a better chance of being cited.
• Account managers who can answer "what are you doing for our AI visibility?" without hesitating.
• Weekly AIO brand monitoring, including claimed YouTube and LinkedIn profiles for the client and a Wikipedia audit.

If you are running this across dozens of clients, the repeat work (weekly snapshots, report splits, citation checks) is a natural fit for AI specialists, leaving human capacity for strategy. For how that looks in agency operations, see the page for agencies.

What to do this week

1. Split existing reports into brand and non-brand, for AI visibility and for paid acquisition.
2. Build a 30-prompt set and freeze it for a month.
3. Take dated AIO screenshots of clients' brand queries and check which sources are cited.
4. Plan your first holdout test and reserve budget for it before peak season.