Your weakest AI assistant shows your brand half as often as your strongest

If you check your brand in one AI assistant and stop there, you may be looking at your weakest channel. Across 207 brands tracked in three or more AI assistants, the weakest assistant mentioned the typical brand half as often as the strongest one. And which assistant is weakest differs from brand to brand.

I've spent most of my career buying attention on Google, Microsoft and LinkedIn. In paid media, I never reported Google and Microsoft as one number, because they behave differently. But that's exactly how most teams report AI visibility: one screenshot from ChatGPT, pasted into a board deck.

Our State of AI visibility report showed that assistants disagree on which competitor wins. This article goes one step further: how far apart they are on your own brand, and what to do with that.

The short answer

• The gap is big. For the typical brand, the weakest assistant mentions it 0.48 times as often as the strongest. That's a 14-point gap.
• Your weakest assistant is your own. Every major assistant was the weakest one for some brands. You can't assume which one it is for yours. You have to check.
• Google's assistants stick together. Gemini, AI Overviews and AI Mode sit within about 4 points of each other. The other assistants each go their own way.
• So report per assistant, weight them by where your buyers ask, and treat your weakest one as its own job.

What we found

We compared, assistant by assistant, the share of answers that mention each brand. We included every brand with at least 200 answers in each of three or more assistants between 4 October and 2 November 2026. That's 207 brands and more than 1.5 million answers.

• Most brands have a big spread. For 129 of 207 brands, the gap between their best and worst assistant was 10 points or more. For 67, it was 20 points or more.

How far apart the assistants are for the typical brand:

Why the same brand gets different answers

Each assistant decides on its own which pages to search, which sources to trust and how many brands to name. Three differences explain most of the gap:

1. They search different indexes. Google's assistants draw on Google's index, which is why they agree with each other. ChatGPT writes its own searches with its own wording.
2. They trust different sources. One leans on review sites, another on news or forums. If you're strong on one kind of source and absent from another, the assistants that prefer the second kind will skip you.
3. They name different numbers of brands. Some list five options, some list two. A short list squeezes out mid-sized brands first.

What to do about it

Today, most teams handle this in one of three ways:

• One ChatGPT screenshot. It's the assistant everyone knows, so it becomes "our AI visibility". But one assistant can sit far above or far below the others, and a screenshot can't tell you which.
• A blended average. One number across all assistants. It hides the assistant where you're weakest, which is usually where the easiest gains are.
• Checking a few by hand. Someone asks the same question in three assistants once a month. One answer per assistant isn't a measurement.

Step 1: Report each assistant as its own channel

Put one line per assistant in your monthly report, the way paid media reports Google, Meta and LinkedIn separately. Show the trend for each, not just the level.

In 3RD: the Overview has a breakdown per model, and the Leaderboard's Models tab shows your visibility in each assistant side by side with every competitor you track.

Your visibility in each assistant, next to every competitor you track. Demo workspace; brand names blurred.

Step 2: Weight them by where your buyers ask

Not every assistant matters equally to you. If your buyers live in Google, AI Overviews and AI Mode deserve more weight than Copilot. If they're developers, Claude and ChatGPT may matter more. Ask sales which assistants prospects mention, and check your analytics for referral traffic from each.

In 3RD: you choose which assistants to track in Monitoring settings, and every metric can be filtered by model.

Step 3: Read the answers in your weakest assistant

Open the answers from your weakest assistant on your most important questions. For each one, note who it names instead of you and which pages it cites. Then do the same in your strongest assistant. The difference between the two lists of cited pages is usually the explanation.

In 3RD: Conversations shows the raw answer from each model and the pages it cited, for every tracked prompt.

The same question answered by five assistants, and who each one names. Demo workspace; brand names changed.

Step 4: Work on the sources your weakest assistant trusts

Filter your citations to the weakest assistant. Look for the third-party pages it cites when it names competitors and not you. That list is usually short. Pitch those pages first, and write the pages on your own site that answer the questions it's missing you on.

In 3RD: filter the Citations page by model to see the sources one assistant leans on. Outreach ranks third-party pages by how many of your prompts they influence and which competitors they feature.

Step 5: Use your strongest assistant as a reference

Whatever your strongest assistant knows about you, it learned from somewhere. The pages behind your best assistant show you what already works. Make sure those facts also exist on the sources your weakest assistant reads.

What 3RD doesn't do

• It doesn't make the assistants agree. Each one builds its own answer.
• It doesn't tell you which assistant your buyers use. That comes from your sales team and your analytics.
• It doesn't publish or pitch for you. It shows which sources each assistant trusts, so you know where to point the work.

Common mistakes

• Reporting one assistant as "AI visibility". For the typical brand, the gap between the weakest and the strongest assistant is 14 points. One assistant can't stand in for the rest.
• Averaging everything into one number. The average hides the gap you can close.
• Treating Google's assistants as three separate jobs. They move together. Work on them as one.
• Comparing assistants on a single day. Each one varies from day to day. Compare trends over weeks. We cover what's normal in When is a change in AI visibility real?.

About the data

• Source: AI answers to the prompts 3RD customers track, 4 October – 2 November 2026.
• Brands: 207 customer brands with at least 200 answers in each of three or more assistants. Internal, demo and test workspaces are excluded.
• Metric: the share of an assistant's answers that mention the brand. All versions of an assistant count as one. Brands track different model versions of the same assistant, so we don't rank the assistants against each other.
• Gaps: the typical gap between two assistants is the median absolute difference across brands tracked in both (40 or more brands per pair).
• Caveats: customers choose which assistants they track, so the mix varies by brand. The sample leans Nordic. Assistants change models over time. We report medians across brands unless we say otherwise, and every number is aggregated and anonymized.

FAQ

Which AI assistant matters most for brand visibility? The one your buyers use. No single assistant stands in for the others: for the typical brand, the weakest assistant mentions it half as often as the strongest.

Is ChatGPT a good proxy for AI visibility overall? Not on its own, and no single assistant is. For the typical brand, the weakest assistant mentions it half as often as the strongest, and which one is weakest varies from brand to brand.

Why is my brand visible in Gemini but not in ChatGPT? They search different indexes and trust different sources. Compare the pages each one cites for the same question to see where the gap comes from.

Do Gemini and Google AI Overviews show the same brands? Mostly. The typical gap between them is under 4 points, compared with 15 points between Copilot and Gemini.

How many assistants should I track? At least the ones your buyers use, plus one from Google and ChatGPT. Three to five covers most teams.

Next step

This week, take your five most important buyer questions and read the answers in your strongest and your weakest assistant. Write down the pages each one cites. The difference is your to-do list.

If you want to see every assistant side by side, with the sources behind each answer, start for free and open the Models tab.

Written by

Magnus Windeløv

Head of Marketing

Magnus leads marketing at 3RD. He has spent his career in demand generation, growth and SEO for B2B software companies and consumer brands.

View profileLinkedIn

Reviewed by

Isabella Willumsen

Solution Architect, AI/GEO Specialist

Isabella is a senior SEO specialist and Solution Architect at 3RD. She has worked in SEO and digital marketing since 2015, across travel, healthcare and multi-market brands.

View profileLinkedIn