The numbers behind generative engine optimization in 2026: what our own research across 1,200 buyer moments found about how ChatGPT, Gemini, Claude, and Perplexity form brand recommendations.
We mapped 1,200 life and business moments by hand and ran each as a natural conversation across all four assistants. Roughly 60 percent already had a default winner, one brand recommended first, most of the time. The full methodology is in our Life Moments Report.
Most AI answers name two to four brands. In categories with ten or more credible players, that means the majority of brands are effectively invisible in a given moment. Presence is a scarce resource, which is exactly why the discipline exists.
Adding qualifiers like budget, team size, or industry changed the recommended set in a substantial share of our runs. Brands that win the generic phrasing frequently lose the qualified variant, and the qualified conversations carry the higher purchase intent.
ChatGPT, Gemini, Claude, and Perplexity overlapped far less than most teams assume. Winning one platform does not imply winning the others, which is why single-assistant measurement produces a misleading picture.
Two implications. First, the open 40 percent of moments is the cheap territory: no default has formed, and early consistent signals can establish one. Second, measurement must be repeated and multi-assistant, because single runs are anecdotes. Book a demo and we will show your category's numbers live.
In our study of 1,200 moments run across ChatGPT, Gemini, Claude, and Perplexity, roughly 60 percent had one brand recommended first in a majority of runs.
Most commercial answers name two to four brands, which makes inclusion scarce and position valuable.
From Aethon's own research: 1,200 hand-mapped moments run repeatedly across the four major assistants, published in the Life Moments Report.