Posted by I get very few emails from my own newsletter list. People read it, use what’s useful, and never write back to tell me what they’re stuck on. For years I read that as a quiet, well behaved audience. Then I started thinking about what that quiet might mean for my AI visibility.
It’s an easy quiet to misread. I assumed silence meant satisfaction, when it might just as easily mean the question went somewhere else entirely.
If somebody wants help with something I teach, and doesn’t email me, where do they go instead? Increasingly it isn’t a search engine. It’s an AI assistant, open on their phone, answering their plain question the way a friend once might have.
Picture someone typing a plain question into an assistant on their phone, maybe waiting for a coffee or standing in a queue. They get an answer in seconds, from whichever source the assistant happens to trust, long before they’d ever think to search for your name directly.
That raises an uncomfortable question for anyone running an online business. When someone asks AI about the thing you teach, does your business get a mention? Or does the answer come from somewhere else entirely, and you never find out?
That’s what I want to look at today: a simple test you can run on your own AI visibility, in under an hour.
The Test That Tells You The Truth
Here’s the test. Write down ten questions a stranger might type into an AI assistant if they were looking for exactly what you offer. Don’t mention your name or your website in any of them.
Not “what does Jim Donnelly sell”. Something closer to “how do I start an online business with no experience” or “what’s a good way to build an email list from nothing”.
Split your ten into three honest groups. A few best questions, where someone wants the top option. A few problem questions, where someone needs help with something specific. A few comparison questions, where someone is weighing up two routes forward.
A best question might be something like “what’s the best way to build an online income from scratch”. A problem question might be “why isn’t my email list growing”. A comparison question might be “should I start with affiliate marketing or build my own product first”.
Write all ten down before you open anything. If you draft them one at a time while already testing, you’ll unconsciously soften them to match answers you’ve already seen.
Then open an AI assistant and ask it every question, one at a time. Write down whether you’re mentioned, who is mentioned instead of you, and where that answer seems to be drawn from.
Don’t be discouraged if you don’t show up once. That’s the whole point of running the test. You’ve just found out something today you didn’t know yesterday.
Most people who run this test are surprised twice. They’re surprised by how often they’re missing, and by how often the mention that does appear is thin, out of date, or simply wrong.
Why Your Own Website Rarely Fixes It Alone
The instinct, once you see the gap, is to write another page about yourself. Add a paragraph, publish a post, hope the next crawl picks it up.
It rarely works that cleanly. AI assistants aren’t only reading your own website. They’re drawing on reviews, forum answers, other people’s articles, and anywhere else your name or your subject happens to come up.
A business might publish ten new pages this month and see no difference in how an assistant answers. That’s because none of those pages are the kind of thing an assistant tends to draw from when it builds an answer.
I’ve covered some of this ground before, in other articles I’ve shared recently, but it’s worth repeating here. It changes what fixing the problem looks like in practice.
Being mentioned somewhere other than your own site, by somebody else, often carries more weight than another paragraph you wrote about yourself. A genuine comment on someone else’s forum thread, or a mention in somebody else’s comparison article, does more for your AI visibility than a tenth page on your own site ever will.
Think of it as reputation you don’t control directly, built in other people’s words rather than your own. That’s precisely why it tends to carry more weight.
What To Do With What You Find
Once you know which questions leave you out, look at who gets mentioned instead. Not to copy them, but to notice where they turn up.
An industry article. A genuine review somebody left unprompted. A forum thread where their name came up on its own. A comparison page nobody paid for.
That tells you where to spend your time next, and it usually isn’t another page on your own website. It’s a genuine, useful comment left on somebody else’s. A guest piece written for someone else’s audience. A mention somebody chooses to write because what you did was worth mentioning.
If three of your ten questions keep surfacing a particular forum, that’s worth joining properly rather than lurking in. If a specific comparison site keeps coming up, a short, honest pitch to be included there will do more than a month of blog posts.
Being mentioned somewhere other than your own site, by somebody else, often carries more weight than another paragraph you wrote about yourself.
Run the same ten questions again in a month. Not because the result will have transformed overnight, but because watching it shift, even slightly, tells you the effort is landing somewhere.
Three months of that pattern, repeated honestly, tells you more than any single snapshot ever could.
A Test Worth Running Regularly, Not Once
I’ve spent a long career trusting feedback I could see straight away. In a shop, if the shelves weren’t full, you knew immediately. In this business, a subscriber can simply go quiet and you’re left guessing.
If the shelves in a supermarket are empty, everyone from the customer to the regional manager knows within the hour. Online, there’s no shelf to check, which is exactly why a test like this matters.
Nobody tells you they didn’t find you. That’s the discipline I had in the physical businesses and lost for a few years online, and it’s the thing I’m getting back now.
AI visibility is another version of that same problem, dressed up in newer clothes. You just don’t show up in an answer you never saw being given.
A customer who once might have emailed you a question now asks an assistant instead, and you never see the conversation happen at all.
Running a test like this doesn’t fix that overnight, and it shouldn’t be treated as a one-off chore to tick off and forget. It gives you a clear, honest number to track, instead of a feeling you can’t quite put your finger on.
I’ll be running mine again next month, on the same ten questions, to see whether anything has shifted.
Start With Ten Questions, Not Thirty
You don’t need a research department or a spare afternoon to do this properly. Ten honest questions, asked plainly, will tell you more about your AI visibility than another week spent guessing.
Pick the five questions closest to someone deciding to buy from you, or to join your list.
Start there. The rest can wait.
It’s a small habit, repeated regularly, that most people never bother building at all. That alone puts you ahead of quite a lot of your competition.
Give it a try this week. Write your ten questions down before you do anything else, and see what the answers tell you.
I’m rooting for whatever you find out.
Jim