Your AI visibility looks fantastic, so does everyone else’s

New SISTRIX research raises a question for marketers celebrating their AI visibility. Have they established what their audience wants, or found a convincing way to reinforce their assumptions?

Consider a hypothetical marketing review. The team has published content around the claims the business wants to be known for, tested related questions in AI search and brought the results to management. The brand appears prominently, so the investment appears justified.

Nobody has established whether prospective customers care about those claims.

We’d argue that the mature decision comes before the dashboard. Establish where the business actually stands, even if the findings make an awkward presentation. Choosing measurements that manufacture success offers a short-term reward while leaving the company ignorant of what it needs to change.

That matters when a competitor can choose a different set of questions and arrive at an equally flattering result. Everyone is doing marvellously, and isn’t that convenient.

Choose your questions, collect your prize

In research published this week, SISTRIX compared brands’ visibility relative to competitors in their own selected prompt sets with their visibility in its broader AI Check question pool.

The company reports that brands looked better in their own sets in roughly three out of four cases. Its article does not give a separate sample size for that particular comparison.

SISTRIX suggests the difference to selection bias as companies choose questions close to their own brands. The finding doesn’t establish deliberate manipulation or prove those questions are irrelevant to customers. Its broader question pool is also a benchmark selected by SISTRIX, rather than a complete record of customer demand.

A specialist supplier has good reason to track specific purchasing questions. We wouldn’t recommend chasing a broader score merely because someone has labelled it independent. But the business needs evidence that its chosen questions matter to buyers, and a messaging document approved by the leadership team isn’t that evidence.

Suppose a business wants to own claim X. It publishes articles explaining X, commissions executive commentary about X and measures whether AI mentions the company when asked about X.

The resulting visibility can be real, but the conclusion that the content is doing its job can still be wrong.
If nobody has established whether customers care about X, the team has measured how successfully it talks about itself. Its competitor can do exactly the same thing with a different claim. Neither needs to understand the audience to produce a reassuring report.

The target audience trusted the brand

We’ve encountered a version of this problem in research focused on perceived buyers within target customer organisations.

The research found that those individuals trusted the company, with its heritage contributing to that trust. The response was to double down on heritage and trust in the messaging.

The trust was real. The conclusions drawn from it were incomplete.

The work hadn’t established who else influenced purchasing or what mattered to them. Nor had it adequately explored criteria such as product performance. Individuals valued the company’s history, and that finding became the justification for telling them more about it.

Meanwhile, the business wasn’t performing as well as competitors.

The research couldn’t explain that disconnect. It hadn’t established whether trust distinguished the company from other suppliers, or whether buyers trusted several businesses and choose between them on performance.

Trust is a lovely thing, but it isn’t exclusive. A competitor can be trusted too, while selling a product that does the job better. In most cases, the performance of trust in influencing purchasing decisions hits a glass ceiling.

There is a level were trust in a brand becomes acceptable, then every company above that line is differentiated by other factors. You could be significantly more trustworthy than another company, but if that company is trustworthy enough, it starts to matter less.

We can’t attribute the commercial performance gap to the messaging from this example. We can say the research left important purchasing questions unanswered, while encouraging the business to reinforce a claim it already liked.

Before writing another article about claim X

We’d start by examining the audience profiles beneath that plan.

Who built them, and what evidence did they use? Do they describe what buyers want to achieve, or what the company wants buyers to appreciate about its products?

If the profiles start from internal assumptions, those assumptions shape the content and the questions used to measure it. The company can work diligently through the entire exercise without ever challenging its starting position.

We’d commission an independent agency to build the profiles afresh, using a deliberately limited brief: what the company does, the types of organisations it sells to and who has traditionally purchased from it. The agency should investigate whether that buying group and its motivations have changed.

Previous profiles and marketing messaging should wait until the new work is complete. Otherwise, the business risks paying an agency to return its own assumptions in a different font.

Then compare the findings. Establish which assumptions still hold and which need replacing.

Being external doesn’t make an agency correct. It still needs evidence for its conclusions about buyers’ motivations, and people have to examine that evidence. An AI visibility tool cannot establish why a customer chooses one supplier over another by counting mentions.

The findings must also change what gets published. If buyers care about product performance, an article should help them understand and assess it. If an internal approver replaces that discussion with another paragraph about the company’s heritage, the research has changed nothing. The content is still serving the person signing it off.

This can require changes to messaging and brand positioning as well as individual articles. Commissioning independent research means accepting that it might challenge something senior management has already approved.

Marketing teams have a responsibility to question the results they present. Senior executives have a responsibility to avoid making reassurance the price of approval.

At the next review, we’d leave the visibility score on screen and ask for the evidence that customers care about the questions behind it. That ought to be a slide the team can produce.