AI assistants don't agree about your brand. Plan for that.

On 23 October 2026 Forbes Agency Council published a post by Tessar Napitupulu, founder and CEO of the generative engine optimisation firm Arfadia. Its argument is that AI assistants often recommend different brands for the same question, so a single blended "AI visibility" score hides where you are being missed. The author recommends reporting each assistant separately.

What numbers does it cite?

The post cites several studies. I have not checked the originals, so treat the figures as the author's. They include BrightEdge data that ChatGPT, Google AI Overviews and Google AI Mode disagreed on brand recommendations for 61.9% of queries, with only 17% returning the same brands on all three. It also cites Ahrefs finding that AI Overviews and AI Mode cited the same web addresses 13.7% of the time.

Two cautions. This is a contributor opinion piece, and the author's agency sells the citation tracking it recommends. The agency's own client examples and internal test are self-reported and unverified.

My take

This is opinion. The direction is right even if the exact numbers are soft. Many founders ask one assistant one question and either panic or relax. If the assistants really do give different answers, one check proves nothing.

I would not buy a single score. I would do something cheaper and more useful: ask the assistants what a buyer would ask, and write down what each says about you. The gap between the answers is your to-do list.

The fix is old-fashioned brand work. Assistants describe a brand from what is written about it. If your own site, your profiles and a few trusted third-party pages say the same clear thing, you give every assistant the same raw material. That is the Clarify and Connect work in the Brand Growth OS™: decide the sentence, then repeat it where buyers and machines look.

You cannot control which assistant answers. You can control whether they all have the same clear sentence to work from.

A one-afternoon test

• Write five buyer questions. The kind a real customer would type, for example "who helps [your type of client] with [your problem]?"
• Put them to each assistant you care about. For most businesses that means ChatGPT, Google, Perplexity and Gemini. Record whether you are named, how you are described and which sources are cited.
• Fix the biggest gap first. If you are described wrongly, correct your own pages. If you are missing, make sure a page on your site states plainly who you are and what you do.
• Repeat monthly. Answers move, so a single check is a snapshot.

The AI Visibility Check walks through the questions, and the AI Mentions Tracker gives you a one-page record to keep each month. If you want a view of where your brand stands overall, run the Brand Growth Audit™.

Sources

Forbes Agency Council, 23 October 2026. Figures attributed to BrightEdge and Ahrefs are as cited there and not independently verified by me. Opinion in this piece is Mike Bruce's. See all brand voices · About Mike Bruce.