When your name isn't in the question.

Leading the witness

Asking an assistant about yourself by name is like asking a witness, "Did you see the man in the blue jacket?" You've put the answer in the question. Of course it has something to say about you. You told it to.

Buyers ask open questions. "Who are the best accountants for small construction firms in Brisbane?" "What should I look for in a commercial fit-out company?" "Is it worth paying for a specialist recruitment agency for engineering roles?" Your name isn't in any of them. Whether it appears in the answer depends on how clearly the assistant understands who you are, who you're for and why someone would choose you.

Visibility is being described when someone asks about you. Being found is being named when someone asks about their problem.

Which stages this tests

In the seven stages we set out at the start of this series, this is mostly about two of them. Found asks, "Who shows up when I search, or ask AI?" Answered asks, "Do they answer what I actually need?" AI assistants now sit right across both. They decide who gets named, then they summarise what those brands offer, often including price, fit and drawbacks.

That second part matters. Even if you're named, the assistant might describe you in a way that cuts you. "A larger firm, typically more expensive" might be fine for one buyer and fatal for another. "Limited information available about pricing" is the assistant telling the money that you weren't willing to say.

How to ask like a buyer

A fair test needs fair questions. Here's how we write them.

Start from the buyer, not the brand

Pick one buyer type and one stage. The champion at Found asks different questions from the money at Answered. Write the questions in their words, about their situation. If you need ideas, look at the questions your sales team hears on first calls, and the search terms that bring people to your site.

Write at least five per moment

One question is an anecdote. Assistants vary their answers, and small changes in wording can change who gets named. Five or more questions for each moment gives you a pattern you can trust. Across all six buyer types and seven stages, that adds up quickly, which is why our own testing asks 200 or more questions for a single brand.

Ask more than one assistant

ChatGPT, Gemini, Perplexity and Google's answers don't always agree, and your buyers don't all use the same one. Ask each question of each. Note who gets named, in what order, and what is said about them.

Keep your name out, and your hopes too

Don't add clues like your suburb if the buyer wouldn't, or the exact phrase from your tagline. Ask the question the way a stranger would. If you only appear when you hint at yourself, you don't appear.

Reading the answers

Once you have the answers, three things are worth recording for each question.

Were you named? Yes or no, and where in the list. Being fifth of five is different from being first.

Who was named instead? These are your real rivals in the buyer's eyes, and they may not be the ones you'd list. Sometimes a brand you've never thought of turns up again and again.

What was said, and where did it come from? Look at how each brand is described, and if the assistant cites sources, look at those. Often the description comes from a directory listing, an old article or a review site rather than your own website. That tells you where the gap is.

Why assistants skip brands

When we see a brand left out, the reasons are usually ordinary. The website talks in broad claims rather than plain answers, so there's little for an assistant to quote. Prices, timelines and limits are missing, so the assistant can't answer the money. The brand describes itself differently in different places, so the assistant isn't confident what it does. Or the brand simply hasn't published answers to the questions buyers actually ask.

None of these is mysterious, and none needs a trick. Clear, consistent, specific pages that answer real questions tend to get understood, by people and by machines.