SECTION 01

The new front door

There is no page two. There is no position eleven. You are in the shortlist, or you do not exist.

It is Friday evening. Someone in your city is cooking duck for four people. Five years ago they typed “wine with duck” into Google and scrolled through ten blue links, three ads and a map. Tonight they open ChatGPT and write: “I'm cooking duck breast with cherries — recommend a wine around $40 I can actually find.” Ten seconds later they have an answer. Not a list of websites — an answer. Three or four specific bottles, each with a reason.

Your wine was in that answer, or it was not. That is the whole story of this report.

The scale of this shift is no longer a prediction. Nearly half of AI users now start product research on an AI platform — 46%, up from 25% just two years ago — while starting on traditional search fell from 43% to 24% in the same period. Meanwhile roughly 60% of Google searches now end without a click to any website at all. The old front door is quietly closing while a new one opens.

Here is the uncomfortable part, borrowed from your neighbours in hospitality. The first industry-wide study of AI recommendations found that 83% of restaurant locations never appear in AI-generated recommendations at all — even though 86% of them are on Google. And when an assistant answers, it names only three to five businesses, then stops. There is no page two. There is no position eleven. You are in the shortlist, or you do not exist.

Now the good news, and it is genuinely good. Almost nobody in wine is working on this yet. Your competitors are still arguing about Instagram reels. The wineries and wine shops that make themselves understandable to these machines in the next year will occupy shortlist seats that are very hard to take back — because assistants keep recommending what has already worked. And the visitors who do arrive from AI answers convert around 4.4 times better than search visitors, because they arrive pre-sold, holding a recommendation that already matched their dinner, their budget and their taste.

This report does four things: it explains what these machines actually are (in plain language, no computer science), shows you exactly how they choose which wines and wineries to name, gives you a five-minute test to see your own visibility, and ends with a 30-day plan you can run with the team you already have. No agency required.