Three passes.
One briefing.
Every SPC cycle runs the same three passes — ingest, model, brief. The pipeline is boring on purpose. The judgment lives in the model layer and the editorial sense of the brief.
Connect
Every feed arrives through a consent-based data share — reviews, booking, PMS, guest-feedback. Your team authorizes the scope, you send it, we analyze it. Every record is de-duplicated, translated if needed, and stamped with a canonical ID. SPC does not scrape third-party platforms; no back channels.
- Sourcing modelConsent-based share
- Feed typesReviews · PMS · booking · direct
- Locales17
- De-dup rate~11%
Analyze
Each review is decomposed into sentiment, sub-aspects, urgency, and language flags. Sub-aspects map to operational owners — not star ratings. A "pillow firmness" complaint resolves to housekeeping, not a rollup.
- Sub-aspects tracked1,240
- Sentiment accuracy95.4%
- Urgency precision91.2%
- Owner assignmentautomatic
Act
Signals are composed into a briefing: what changed, what it will cost, what to do today. Urgent items are surfaced with rationale, not just a score. The editorial voice is deliberate — a board-ready paragraph beats a chart.
- Median lead time21 days
- SLA on critical14 days
- Draft responsesAI-suggested
- Briefing cadences5
Signals extracted
Where the signal comes from — and how often.
Coverage and freshness by channel, as of the latest cycle. Direct-feedback streams (post-stay email, in-room tablets, front-desk chat) are normalized into the same schema and carry higher weight in the urgency model.
Source A
Source B
Source C
Source D
Urgency, detected — a 14-day walk-through.
The same signal landed three times in two weeks before it made the dashboard as critical. Standard BI would have missed the first two; SPC's pattern layer keeps a rolling window on topic × location and escalates on the third independent mention.