Intervention

Invisible to AI, invisible to the market. The risk that marketing managers can no longer ignore

The conversational search does not return lists of results. It picks a winner. And companies excluded from the answers do not get a second chance

by Ale Agostini*

23 April 2026

3' min read

Translated by AI

Versione italiana

3' min read

Translated by AI

Versione italiana

Something structural has changed in the way people make purchasing decisions. It is not an algorithmic update, it is not a new feature of Google or Instagram. It is a disruption of the model that has governed information research from 2001 onwards.

The consumer, increasingly attentive to the use of his time, has stopped selecting options from an endless list of links. Today he prepares and sometimes delegates the decision to language models such as ChatGPT, Gemini or Claude, demanding immediate and definitive answers. And those models save a lot of time for the user who, instead of scrolling through five or ten results, reads one or maybe two answers.

The magnitude of this change is momentous. A drop in ranking on classic Google represents a technical problem. Exclusion from AI responses decrees immediate irrelevance and in the medium term a commercial extinction.

The power of choice has changed hands

The many changes in search and digital over the last twenty years changed the technical rules, but left a fundamental balance intact: the user searched, the engine returned links, the user invested time to choose, but retained the power of choice. The Conversational AI broke that balance. The algorithm suggests few options and influences the final decision instead of the consumer. A recent research conducted for a client on a sample of 1,000 B2B buyers showed that, for 35 per cent of the respondents, AI suggestions actually affect the final purchase decision. A brand excluded from the AI response is not on the second page. It does not exist on the radar of the potential customer and questions the relevance even of customers who know it.

SEO is not obsolete. Immobility is

One of the most frequently asked questions by companies is whether the SEO investments accumulated over the years still make sense. The answer is yes, as SEO and GEO have different areas of contact, but with one condition. Those with a solid SEO architecture possess the best ground to impose their identity on AI. The real risk is in standing still and choosing not to decide and waiting at the window.

Generative Engine Optimisation complements SEO and extends it to a new level. Authoritative content, the technical structure of the site, and the digital reputation built up over time become the signals that generative models use to decide who to cite. Those who have already invested in this asset have an advantage. Those who stop now risk being left with nothing.

The 'One-Shot Brand Reputation'

In the conversational search, there is only the answer, and there is no second page of results in which to hope to be found. The user receives one recommendation, at most two or three. Stop, end of transmission. An omission on the part of the AI in favour of competitors results in an immediate loss of relevance and eventually market share.

This resets the margin for error to zero. And it demands a concrete and decisive change of approach from top management.

What's a marketing director to do now?

The priority is to understand where and how one's brand appears in the responses of the main AI models today, and where it does not appear when it should be there. This is not a one-off analysis: it is an ongoing monitoring, as those models update and their reference sources change.

The second level is data consistency at the brand level. The AI, which responds even when it does not have all the info, aggregates information from dozens of sources: official website, press articles, reviews, social profiles, industry directories. If these sources contradict each other, or if some are incomplete, the model fills in the gaps with approximations. With bad consequences: brand integrity on AI gets the wrong end of the stick, the brand loses control over its own narrative just when someone is interested in it.

The third level is the production of content designed to answer specific questions, not to rank for generic keywords. Generative models look for answers and snippets of content that get to the point, not long, boring pages of corporateese that no one has the time or inclination to read. A brand that structures its content around real questions from its audience increases the likelihood of being the answer and authoritative source in its industry.

What is at stake is not Google positioning. It is presence or absence at the exact moment when a potential customer asks an AI what to buy, what service to choose, which company to trust. Those who face this new challenge today are ahead of the competition, those who are asleep have already lost precious ground.

CEO of AvantGrade, author of the book "Being the Answer"