AI Recommendation Openness

The score describes the recommendation pattern in one dated sample. It does not tell you how many businesses operate in the city, how much customers spend, or whether entering the market will be easy. “Open” means the assistants spread their recommendations across more names in our answers—not that the real-world market has spare capacity.

The score combines three equally weighted observations. First, how often did the most-named business appear? Second, how many repeat names did ChatGPT, Gemini, and Perplexity share? Third, what share of businesses appeared only once? A market scores higher when there is no dominant repeat name, the assistants share fewer names, and one-off mentions make up more of the list.

Averaged across five metros, dental practices scored 84.6, law firms 68.4, HVAC companies 66.4, med spas 63.2, roofing companies 62.0, and plumbing companies 60.2. In this sample, dental recommendations were spread across more names; plumbing recommendations repeated a smaller set more often. The gap between trades was wider than the city range inside any one trade.

The limits matter. Each market has only 18 answers, AI responses change, and the equal weights are our stated design choice rather than a scientifically optimized formula. The score is useful for comparing these sampled recommendation patterns, not for predicting revenue or declaring a winner.

The formula, thresholds, and limits are published in full on the methodology page, and every score appears beside its raw inputs. A metric nobody can recompute is marketing; this one is meant to be checked.