The world’s top 500 incentive hotels.

The exact formulas

Every pillar is reproducible. Computed pillars blend a published editorial base 50/50 with a live data signal — so a base never compounds, and every score decomposes cleanly to its inputs.

Guest Experience now triangulates two independent rating sources— Google and TripAdvisor — each Bayesian-adjusted on its own prior, then combined by review weight so neither source dominates on volume alone. TripAdvisor’s structured Location and Value subratings harden the Setting and Value pillars, and the pooled review count across both sources drives the base decay, so two sources agreeing raises data-confidence faster than one source ever could.

Editorial bases — explicit, and shrinking

Where live data is thin, a pillar leans on an editorial base — but a base is a documented scorecard, not an opinion, and it is a Bayesian prior that shrinks as data lands. Its weight in the blend decays from 50% toward a 5% floor as the computed signal gains sample size:

baseWeight(evidence) = max( 5%, 50% × 200 ÷ (200 + evidence) )
→ 40 reviews: base ≈ 42% · 300: ≈ 30% · 2,000: ≈ 9% · 9,000: ≈ 5%

Each base scores against published criteria:

Where the scores come from

Every pillar is labeled COMPUTED or EDITORIAL on each venue’s page, per venue, based on the data behind it. Value computes from live median nightly rates (Google Hotels) via the published quality-adjusted-affordability formula — value = 60 + (GE_base − 85) × 1.2 − 22 × ln(rate / $500), blended 50/50 with the editorial base. Guest Experience computes from Google ratings with a Bayesian prior — a 4.45★ prior weighted as 300 reviews, then 76 + (adjusted − 4.45) × 40, blended 50/50 with the editorial base. Each venue’s page shows the blend math with its own numbers.

Group Readiness stays editorial, anchored to property facts — room count, indoor meeting space, realistic group capacity, and buyout culture — shown on every venue page. Setting and Dining & Wellness are editorial judgments against the published criteria. Every change lands in an append-only readings ledger and rankings recompute after every data pass, the same way the Destination Index™ publishes which pillars are live, computed, or editorial. Venues may submit verified data for the next edition.