Comparable measurements, not random screenshots
A single chat screenshot says nothing about a trend — model answers vary between sessions. GeoFolks measures under a fixed protocol: the same intents, the same models, controlled repetitions, and timestamps.
Only that comparability lets you tell a real visibility change apart from the natural variance of answers.
Available alert rules
Rules can react to a score or confidence drop, recommendation or source loss, a new competitor, run failure, stale canonical data, and an overdue critical action.
A factual-error detector exists in the alert layer, but the real collector currently supplies no fact comparison. We therefore do not present it as an available monitoring capability.