Decision Support for Tourism Risk and Heritage Management: A Case Study of Ha Long Bay

Tourism firms at heritage destinations must manage operational risks and protect the resources on which their businesses depend. Online reviews could inform managerial follow-up, but sparse mentions and high ratings may limit their value. This study examines these limits through a transparent business analytics pipeline applied to 240 English-language TripAdvisor reviews of 30 cruise, hotel and restaurant listings in Ha Long Bay, Vietnam, published between 2024 and 2026. Reviews were coded for risk-responsive practice, heritage-conserving practice, mention valence and recommendation or revisit language, termed advocacy. Group comparisons, business-clustered regression with small-sample corrections and repeated cross-validation holding out whole businesses linked coded signals to ratings and advocacy. Risk and heritage signals appeared in 9.6% and 2.9% of reviews, respectively. Negative risk mentions coincided with ratings 1.81 stars lower descriptively, although the adjusted estimate was imprecise. Positive risk mentions showed no clear rating advantage, and heritage mentions were insufficient for reliable inference. Each additional star was associated with 5.85 times the odds of advocacy, yet the rating-and-segment model discriminated advocacy poorly in held-out businesses; adding review length raised the area under the curve from 0.52 to 0.63. Segment moderation was inconclusive. The contribution is an auditable monitoring routine that prioritises investigation while retaining explicit denominators, uncertainty and checks beyond reviews. It supports managerial attention allocation without establishing causal effects, future recommendation behaviour or verified safety and conservation performance.