Try this. Ask ChatGPT for the best running shoe for a first marathon.
Then keep going. Tell it your budget. Tell it you have flat feet and train on pavement through a New York winter. By the fourth question, the brand it named first has often disappeared.
AIVO, the firm Tim de Rosen founded, ran simulated buying journeys exactly like this. The brand appeared in the AI’s early answers 96 percent of the time. It was the final recommendation 13 percent of the time.
And yet an entire industry, nearly half a billion dollars of venture money in less than two years, has been built to count the first answer: mentions, citations, share of voice. We have seen this movie before. Impressions weren’t sales, and clicks weren’t preference.
What decides the final answer is what has always decided it: the situation in which demand forms. I call these demand spaces, and most companies can’t see them because they are looking at their categories.
It gets more interesting. Walmart has just opened its doors to outside shopping agents, on Walmart’s terms. Soon your own agent will be negotiating with your customer’s agent. Then the question is no longer only whether you get chosen. It’s what you are willing to give up to get chosen. Most brands haven’t told their agent.