ALBUQUERQUE, N.M. — Tokino has published a home-service AI study finding that the business being checked was absent from nearly six in ten sampled answers. The tracked business appeared in 312 of 775 returned answers and was omitted from 463.
For a contractor, the question is practical: when someone asks AI who to call in their city, does their business appear? The public study lets owners examine that question across assistants and trade groups, then check their own business.
The study covers eight selected US home-service businesses across five trades and eight cities. It includes plumbing, heating and air conditioning, roofing, garage doors and pest control. Results compare ChatGPT, Claude, Perplexity and Google AI Mode. Each assistant's result pools the businesses checked; it is not a ranking of the assistants.
For each selected business, Tokino asked five questions about its trade and city, repeating each question five times on each assistant. The questions asked for local recommendations without naming the business being tracked. Sampling took place on November 1, 2026, UTC. Of 800 planned checks, 25 were unavailable and excluded from the results.
The study is a snapshot of selected reference businesses, rather than a random sample of US contractors. Its pooled rate cannot predict an individual business's result. ChatGPT answers were collected from its app; the other assistants were sampled through software connections, which can differ from consumer apps.
The published page includes results by assistant and trade group, the questions used, and the method and limits. Individual business names, answer text and source links are withheld to keep firms out of the results. The study measures neither improvement over time nor the effect of Tokino's services.
Read the full findings, comparisons and method at Tokino.