Visibility in AI answers: how brands organise their presence

FrenchWeb reported on 9 November 2026 that large companies are working on how they appear in AI-generated answers, based on an interview with GetMint about customer needs, measurement, sources and coordination between teams. The report does not specify which companies were studied, what measurement methods are used, or what results were observed. What it does make clear is that visibility in AI answers is being treated as an organisational question, not only as a content question.

From ranking to being cited

For two decades, organic acquisition rested on a fairly stable loop: publish content, earn links, rank in a search results page, capture the click. Assistant-style answers change the last step. The user may never reach the site, and the brand may appear only as a source inside a synthesised response.

That shifts the unit of work. Teams can no longer judge a page solely by its position and click-through rate; they need to know whether the underlying material is retrievable, attributable and current. Established context: answer engines tend to rely on pages that are clearly structured, factually consistent and easy to quote, which overlaps heavily with what search engines already reward.

What the report leaves open

FrenchWeb mentions measurement without saying how it is done. This is the weak point of the topic today. There is no universal dashboard for assistant visibility, and plausible proxies differ in what they capture: referral traffic from AI surfaces, branded search volume, mentions in answers observed manually, or share of citations in a sample of prompts. Each has limits, and the report does not state which approach the companies discussed actually use.

The same applies to sources. A brand can influence what it publishes on its own site, less so what third parties, marketplaces, review sites and forums say about it. If assistants lean on those external sources, a visibility strategy has to include them, which usually means work outside the marketing department.

Coordination is the third theme the report raises without detail. Content, SEO, PR, product documentation and customer support all produce material that can end up cited. When those teams do not share a view of what the company wants to be known for, the answer engines assemble an incoherent picture.

What a team can check this week

• Pick twenty questions real prospects ask before buying, run them through two or three assistants, and record which domains are cited and whether your brand appears.
• Audit the pages most likely to be quoted: prices, specifications, comparisons, definitions. Check that facts are stated plainly and dated.

A useful starting point is to define, in one page, the handful of claims the company wants repeated and who owns each one internally. Without that, measurement produces numbers nobody can act on. The FrenchWeb report identifies the right questions; the answers remain to be built case by case.