Ask Stay Intelligence Model

How AskSIM compiles the true service score

This is the live method on the Las Vegas board — not a brochure. The percentage on each house is Stay Truth: the journey pattern of guest comments on arrival, room, dining, and every venue they used. A failed room cannot be averaged away by a greeting. Market Voice stays beside it and is never mixed in.

AskSIM means Ask Stay Intelligence Model: we ask the stay — guest comments and surveys — and model the true service score.
Other ranking systems publish false information. Booking uses /10, TripAdvisor /5, Expedia /10 or /5. A mean is not percent delighted. Harris 80% Top-2 is not GEM 80 points. Mixing them is invalid math.
The Guest Review does not buy a star, and a hotel cannot buy one either.
Stay Truth is the journey pattern of the touch points we already ask: arrival, room, dining, and every venue the guest actually used. Top-box only. A 90 greeting and a 40 dirty room is not “65” — the failed must-be stage caps the mark. Unused spa or casino scores do not enter. Unfixed problems multiply the mark down. Return intent is shown beside Stay Truth; it is not averaged in.
On the public board, classified guest comments fill those same touch points. Thin comment samples shrink toward Market Voice so two quotes cannot tank a house. When the stay sample is large enough, the live percentage is the journey-pattern lower bound — not a Booking mean.
Market Voice remains the n-weighted share of guests at 4+ after the correct scale. A 9.8 on /10 is not 100. It is shown beside Stay Truth and is never averaged into it.
We collect those published scores and comments daily with cited web search — not by attacking hotel or OTA systems.
Comments and mentions are stored with a source URL and classified to the department. Unsourced text is discarded.
Properties may subscribe for action plans and named team-member comments. That purchase never changes the public mark.