Four pillars
of revenue intelligence.

Sentiment, in our model, is not a single score. Every review is projected across 10 top-level topics and 1,240 sub-aspects, each with its own ownership, elasticity, and historical baseline. A 0.78 property rating is a summary; the Δ on housekeeping vs. the Δ on location tells you where to act.

The model is calibrated quarterly on human-labeled samples and validated against occupancy, repeat-booking, and ADR trajectories. Accuracy is reported on a per-topic basis — we do not pretend that "location" and "housekeeping" behave the same.

Urgency is not sentiment. A 2-star review complaining about a cold room is medium; a 4-star review describing a sanitation incident is critical. We classify explicitly: health & safety, service failure, pattern-of-three, and reputational. Each carries its own SLA, escalation chain, and suggested response style.

The pattern-of-three rule is the workhorse. Three independent mentions of the same topic at the same property within a 14-day rolling window — even if each is individually mild — becomes critical. The rule catches staffing gaps, equipment failures, and environmental drift that single-review thresholds miss.

Traditional comp-set reporting tells you where you sit. SPC tells you why. If Property Delta gained two pts of sentiment this quarter, the model isolates the topic (housekeeping), the source (34% of their new 5-stars cite it), the inflection point (early February), and the operational move behind it — all before a quarterly STR report lands.

Your comp set is editable. We recommend 4–6 properties by default based on room type, rate, and location; many operators add or remove one based on their read of the market.

A -4% sentiment week can mean your housekeeping fell apart, or it can mean the regional airport had 42 minutes of average delay three days running. SPC estimates both contributions. If the signal is 70% external, you adjust your messaging; if it is 70% internal, you adjust your operations.

We track 38 external signals in the base model. The correlation matrix below shows what the data says for your category, not ours — the numbers come from your property's historical relationship with each signal.