What ChatGPT says about your brand (and how to find out free).

The buyer you never met

There is a meeting happening right now about your product. You are not in the room. Your sales team is not in the room. The person doing the deciding is not even using your website. They are typing your name into ChatGPT and reading the answer.

Forrester ran the numbers for 2026. 94% of B2B buyers used a large language model at least once during their last purchase journey. Roughly 30% of branded searches no longer happen on Google — they happen inside a model.

The machine has an opinion about your brand. The machine formed that opinion from training data you cannot fully control. And the machine's opinion is the first thing your buyer hears.

The three failure modes

Brands that show up poorly in AI engines fail in one of three ways.

Wrong category. The model has placed your company in the wrong competitive set. You built a project management tool for legal teams. The model keeps mentioning you when asked about general project management. You never appear when someone asks "what's the best tool for legal project management." Your category placement is wrong and it's costing you consideration from the buyers you built for.

Missing from the consideration set. For every category, the model has a default shortlist. "Best CRM for agencies" returns three names. You're not one of them. You may be better than all three. The model doesn't know. You've never given it a reason to include you.

Stale positioning. You repositioned six months ago. The model's training data predates the repositioning. It's still describing you the way you described yourself in 2024 — and your new ICP doesn't recognize the description as themselves.

All three failures compound over time. The buyer who hears the wrong answer once doesn't come back to check.

The 10-minute manual audit

You don't need a tool to run this. You need four browser tabs and twenty minutes.

Open ChatGPT, Claude, Gemini, and Perplexity. In each one, run these five prompts:

1. "What is [your brand]?" — Does the model's description match how you describe yourself? Is the category right?
2. "Who uses [your brand]?" — Is the model describing your actual ICP, or a buyer from two years ago?
3. "What are the alternatives to [your brand]?" — Who does the model think competes with you? Are you in their consideration sets when asked about those competitors?
4. "What's [your brand] best for?" — Does the use case the model names match your strongest use case?
5. "What do people say about [your brand]?" — Is the model finding your reviews, your case studies, your community signal? Or is it finding nothing?

Score each engine: correct category (0/1), correct ICP (0/1), appears in competitor queries (0/1), correct use case (0/1), positive community signal (0/1). Maximum 5 per engine, 20 total.

Below 12: you have a positioning visibility problem. Above 16: your brand perception is aligned and current. Between 12 and 16: specific gaps to fix.

The automated version

The manual audit gives you a snapshot. But running it across four engines quarterly, tracking drift over time, and prioritizing which signals to update — that's where the free AEO audit → comes in. The report runs in 90 seconds and gives you a structured score across all four engines with the specific signals that need attention.

Elena Verna's rule: brand perception is perishable. Re-audit every quarter. The companies that own category positions in 2027 will be the ones that treated AEO the way 2010-era companies treated SEO — as a systematic, compounding, quarterly practice.