Intent Ambiguity intent_ambiguity
The model measures how unclear the person's goal is, for example "tree service Denver" could mean removal, trimming or stump grinding, and may hedge or cover several readings in the answer.
What you can do: Give each service its own clearly named page, so the model can match you whichever reading of the query it picks.
query.uncertainty (was query.intent in the provisional inventory: Measures how ambiguous the goal is) · also called intent clarity, query intent
Term: search-intent
User Risk user_risk
The model assesses what could go wrong for the person if it recommends badly, such as an unlicensed electrician or an unsafe tree removal, and is more cautious where harm is possible.
What you can do: State licenses, insurance and safety practices on your site, with license numbers where your trade issues them.
evidence.assessment (was query.uncertainty in the provisional inventory: Judges harm risk of a bad recommendation) · also called risk assessment, user safety
Answer Confidence Dependency answer_confidence_dependency
The model judges how much its confidence in the answer depends on how complex the query is; a narrow request like "24-hour locksmith in Boise" can be answered more surely than a vague multi-part one.
query.uncertainty · also called confidence level, response certainty
Term: confidence-score
Evidence Variance evidence_variance
The model measures how much the evidence about a business disagrees across sources, such as different hours or prices on the site and on directories, and lowers trust when it varies.
What you can do: Keep hours, prices, address and phone identical across your site, Business Profile and directory listings.
claims.verification (was query.uncertainty in the provisional inventory: Cross-source disagreement about facts) · also called evidence consistency, data reliability
Checked by Field-by-field audit of your listings