Burson has released research on generative engine optimization that identifies a measurable gap between visibility and believability. The headline finding: brands can appear frequently in AI-generated answers yet still fail to persuade the people who read them.

The study frames this as the central problem of the new search era. Generative engines such as AI Overviews cite sources at scale. Frequency of citation, Burson argues, does not equal trust. A brand can win the mention and lose the argument.

What does the research actually claim?

The core claim is structural. Generative engine optimization has so far optimized for presence — getting named in model outputs. Burson separates that from believability, which it treats as a distinct outcome requiring different work. The firm describes the two as diverging signals that marketers currently conflate.

This is a research and advisory finding, not a statement from any search engine. Google has not confirmed any believability metric. Treat the gap as Burson's analytical claim.

Why does the distinction matter?

If visibility and believability diverge, then ranking-style measurements imported from traditional SEO will mislead. A report showing heavy brand mentions in AI answers may hide weak audience trust in those same mentions. Burson positions believability as the variable that converts AI visibility into commercial outcomes.

What is not said?

The announcement does not publish the sample size, the number of brands studied, or the exact gap measured between the two dimensions. It names no specific generative engines tested. Without those figures, the findings resist independent verification. That absence matters more than the framing does.

No percentages. No dates for data collection. No methodology. The press material gives direction, not magnitude.

What should practitioners watch next?

Watch for the full methodology and quantified results. Until Burson publishes the numbers behind the visibility-versus-believability gap, practitioners should treat the finding as a hypothesis worth testing against their own AI-referral traffic and brand-lift data.