A few years ago, checking how your business showed up in search meant typing your name into Google and scrolling. That still works. But a growing share of people now ask ChatGPT, Perplexity, Google’s AI results, or Copilot to recommend a plumber, a payroll provider, or a project management tool — and those tools answer in a paragraph, naming two or three companies, with no page two.
If your business isn’t one of the names in that paragraph, you don’t appear at all. There’s no equivalent of “ranking eleventh.”
The good news is that you can check this yourself, today, without buying anything. Here’s how I’d go about it.
Step 1: Write down the questions your customers actually ask
Not keywords. Questions, in full sentences, the way a person would type them into a chat window.
If you run a bookkeeping firm in Hamilton, that list looks something like:
- “Who are the best bookkeepers in Hamilton Ontario?”
- “How much should a small business pay for bookkeeping?”
- “What should I look for when hiring a bookkeeper?”
- “Is it worth hiring a bookkeeper or should I use software?”
Ten to twenty of these is enough to start. Mix the ones where someone is ready to hire with the ones where they’re still working out what they need — the second group is usually where you get mentioned first, because the answer is advice rather than a recommendation.
Step 2: Ask them, and write down what comes back
Open ChatGPT, Perplexity, and Google. Ask each question. For each one, record three things in a spreadsheet:
- Were you mentioned?Yes or no.
- Who else was?Every company named, in order.
- What did it cite?Perplexity and Google’s AI results show their sources. ChatGPT often does too. Note the actual URLs.
That third column is the one people skip, and it’s the most useful. It tells you which pages on the internet are feeding the answer about your industry. Those pages are the target.
Step 3: Do it more than once
This is where most informal checks fall down. Ask the same question twice in an hour and you can get two different answers — different companies named, different sources cited. These systems don’t return a fixed result the way a search ranking does.
So a single check tells you very little. Run each question three to five times, on different days, and record every run. What you’re looking for isn’t “am I mentioned” but “how often am I mentioned out of how many attempts.” One mention in fifteen runs and eleven in fifteen are completely different situations, and you can’t tell them apart from a single look.
This is tedious by hand, which is why it’s worth being honest that most people won’t keep it up past the first week. Doing it once properly, writing down the numbers, and repeating it in three months is more useful than doing it casually every day.
Step 4: Look at what got cited, not what got said
Once you have your sources column filled in, sort it. You’ll usually find the same handful of domains over and over: an industry roundup, a directory, a trade publication, a couple of competitor blog posts, and often a Reddit thread.
That list is your actual to-do list. Not “write more blog posts” — get onto the pages that are already being read back to your customers. Sometimes that means asking to be included in a roundup that left you out. Sometimes it means the information about you on a directory is three years old and needs updating. Sometimes a competitor has written the definitive answer to a question in your field and nobody else has bothered.
Step 5: Fix the page they’d land on
If an assistant does name you, someone clicks through. Check what they find.
The pages that get quoted well tend to share a few traits: they answer the question in the first paragraph rather than building to it, they use headings that match how people ask things, and they include specifics — prices, timelines, what’s included, what isn’t. Pages that open with “In today’s fast-paced digital landscape” tend not to get quoted, because there’s nothing in the first hundred words worth quoting.
What this won’t tell you
Two honest limits.
First, you can’t see volume. There’s no equivalent of search volume data for what people ask assistants, so you can’t tell whether a question you’re winning gets asked a thousand times a month or twice.
Second, you can’t prove causation from this alone. If your mention rate goes from 2 in 15 to 9 in 15 over a quarter, something changed — but so did the models themselves, probably more than once. Record what you did and when, so at least you have the timeline. Treat it as a measurement, not a proof.
Where to start
Pick five questions. Ask each one five times across two assistants. Write down the answers. That’s an afternoon, and it will tell you more about your position than any amount of reading about this will.
If you want to go further than a spreadsheet, DaSilva Consulting does this work at a larger scale for clients, running the same method across several hundred questions per brand. But the method above is the whole of it — there’s nothing in the paid version that isn’t in the free version, just more questions and more repetitions.
Tiffany DaSilva runs DaSilva Consulting, an SEO consultancy in Toronto. She teaches SEO at McMaster Continuing Education and AI marketing at Growclass.See More