Your AI Visibility Strategy Is a Dashboard and a Hope

Most of what gets sold as an AI visibility strategy is a dashboard and a hope. I can say that with a straight face because I sell the dashboards. What I got curious about was the hope, so we put it to the test. Over two months we ran 121 buyer questions through two leading assistants at the same moment, in the same category, and compared the companies each one recommended. They agreed on about a fifth of the names. The median was worse, closer to one shared name in nine. One run in five, they had no company in common at all.

Nobody optimizes for that, because nobody can see it. In AI search there is no page two, and no prize for eleventh place. When an assistant leaves you out, it does not rank you lower. It forgets you exist, which is worse and much quieter. There is no line in your analytics for the sale that went to a competitor you were never shown against, on a question you never knew was being asked. That churn is exactly why a whole category of best AI visibility tools now exists, to ask these questions on a schedule instead of guessing from a single screenshot.

Why AI recommendations keep changing

The comforting theory is that a model either likes you or it does not, and once you are in, you are in. We measured that theory 121 times. It did not survive.

The reason is dull and it matters. One assistant answered from memory, from what it absorbed during training. The other could search the live web while it answered. Those are two different competitions. Memory rewards years of reviews, press coverage and category fame, and it changes slowly, on the model maker's schedule rather than yours. Search rewards whatever is published, clear and quotable on the exact question at the moment it is asked, which a small company can influence in weeks. They end up recommending different companies because they are answering slightly different questions. One is asking who is famous for this. The other is asking who has something useful to say about it right now. Reputation is what a model remembers. Relevance is what it can quote today. Confusing the two is how good companies stay invisible.

I ran the obvious objection into the data, because someone always asks it: maybe I only measured "searching" against "not searching." So we split the runs. Where neither assistant searched, the disagreement barely moved. That is too small a slice to wave around as proof, but it points the same way rather than the opposite one. Turning search off did not make the two agree.

What AI assistants actually read before recommending you

When an assistant does search, it usually shows its sources, and the sources are more honest than any sales deck. We logged them across several runs in our own category, on the same eight questions, hours apart. Two things held every time. Video was the most cited source. Community threads were cited in every single run. Almost everything else churned from one run to the next, for identical questions, which is the real reason a one-time check tells you close to nothing.

That has a blunt practical edge. If the assistants keep quoting video and community discussion, those are not a "nice to have," they are where the recommendation is decided. It is also why the more serious tools in this space now watch Reddit directly. Ours flags the threads where a brand comes up, so you can answer as yourself, in the open, rather than find out three weeks late that the conversation happened without you.

If you want the uncomfortable version, here is ours. We build the software that measures exactly this, ran our own brand through eight buyer questions in our own category, and scored a clean zero out of eight. Nothing focuses a product roadmap like being invisible to your own product.

Where your business stands in AI search

There are three honest positions, and each one calls for different work.

You can be remembered but not found. Your reputation carries the assistants that answer from memory, while your current pages give the searching ones nothing to quote. It feels comfortable, and it decays quietly, because reputation is a balance you spend rather than one you top up on demand.

You can be found but not remembered. Your content is winning in the present tense, and you are invisible to anyone whose assistant is not searching. This is most small and mid-sized companies, and it is the better problem to have, because it is the one you can actually move this quarter.

Or you can be in neither, in which case the first useful act is admitting it rather than buying a report that dresses it up. Being remembered and being found are two different achievements, and most brands are failing the one they never measured.

How to measure your AI visibility without kidding yourself

You do not need a tool for a first read. You need an hour and the discipline to do it twice.

Write the ten questions a buyer actually types right before choosing, in their words, not your brochure's. "Reliable payroll provider for a 20-person firm." "Who audits clinics for compliance." Ask each in a fresh chat, and write down three things: whether you are named, who gets named instead of you, and which sites the assistant leaned on. Then ask a second assistant, because by now you know the two rarely agree. Then, the part everyone skips, do it again a few days later. A single result is a snapshot of a moving target. One good screenshot proves nothing except that you own a screenshot.

You can keep this up by hand, the same way you can mow a field with scissors. Technically possible, spiritually punishing, and abandoned by most teams after the first alarming pass. At some point you either automate it or quietly stop and go back to hoping.

What actually moves the answer

None of this is a trick, and none of it is fast. The companies that move are the ones that measure before they start and again afterward, so they can tell which effort changed the answer and which merely felt productive. On the search side that means clear, quotable pages, a short video that answers a real buyer question, and an honest presence in the threads your customers already read. On the memory side it means reviews, press and consistent listings, compounding over months rather than days.

So before you buy anyone's AI visibility strategy, including a version of mine, ask exactly what it measures and how often it measures it. A number you watch beats a dashboard you admire. The whole game is knowing which half of the shortlist you are missing, the half that remembers you or the half that can find you, and then doing something about the one you can still reach.

About the author: Alex Spender is the founder of Citenzo, which measures how often AI assistants recommend a brand and which sources those answers are built on.