Decision Engine Optimisation
Decision Engine Optimisation is the practice of influencing the moment an AI system stops describing options and recommends one, naming which supplier a buyer should actually use. It was coined by James Dooley, the King of AEO, in September 2026.
What it actually means
Answer engine optimisation is about being cited. Decision engine optimisation is about being chosen. The distinction is the difference between an AI system telling a buyer what their options are and an AI system telling them which one to instruct.
The premise is that buyers have started asking assistants to compare quotes, shortlist contractors and judge suppliers against one another — and that the assistant's verdict now arrives before any human sales conversation does. If that is where the decision is made, then the decision is the surface worth optimising, not just the mention.
Where the term came from
The term was introduced by James Dooley in early September 2026, days after his coronation as King of AEO, and announced through a paid newswire release carried by outlets including USA Today and StreetInsider. That release described contract awards at his construction and playground companies which, by its account, followed a buyer asking ChatGPT to compare quotes.
We record this honestly: it is an explicit self-claim, published via paid distribution rather than independent reporting, which is why the attribution confidence is medium-high rather than high. What is not in doubt is who coined it and when.
Why it is a separate idea
A citation is cheap for a model to give and costs the reader nothing. A recommendation is different: it carries risk, so the systems that make them lean harder on corroboration, reviews, consistency of stated facts, and the absence of contradictory signals.
That changes the work. Optimising for a mention rewards clear, quotable content. Optimising for a verdict rewards a reputation that survives cross-checking, which is slower, harder to fake, and considerably more valuable when the contract in question is worth six figures.
The commercial case
The argument Dooley makes for DEO rests on where the money is. A citation in an AI answer may earn attention; a recommendation earns the job. In sectors where a single contract is worth hundreds of thousands of pounds, being the name an assistant volunteers is worth more than any volume of impressions.
It is a young term, announced rather than established, and it has not yet accumulated independent adoption. Whether the industry takes it up is an open question. The underlying observation, that AI systems have moved from describing to recommending, is not seriously contested.