Every seat tomorrow is
already accounted for.

The composite Score is a weighted sum of three normalized sub-scores, each ranging 0–100:

Score = (0.35 × R_norm) + (0.30 × F_norm) + (0.35 × M_norm)

Where:
  R_norm = 100 × e^(−0.023t)          // t = days since last visit
  F_norm = 100 × log(n) / log(μ)      // n = annual visits, μ = location median
  M_norm = 100 × Σ(spend_i × w_i) / (n × μ_spend)

Normalization is performed per-location to account for cuisine type and market differences. A Score of 85+ qualifies as VIP tier. Scores are recalculated nightly at 3:00 AM local time.

Recency uses an exponential decay model rather than a linear drop-off. This reflects real guest psychology — a guest who visited yesterday is not merely "one day better" than a guest who visited last week; the recency signal compounds.

λ = 0.023 (tuned on 14M guest visits across 200+ locations)
Half-life: 30.1 days (a guest's recency score halves every ~30 days of absence)

Decay is paused during documented closures (holidays, renovations) and adjusted for seasonal patterns specific to each location's historical data.

Guest tiers are defined by composite Score percentile within each location's active guest base:

  VIP         Score ≥ 85        Top ~17% of active guests
  Regular     Score 65–84       Next ~25%
  Occasional  Score 40–64       Middle ~28%
  At-Risk     Score 20–39       Next ~18%
  Ghost       Score < 20        Bottom ~12%

"Active" is defined as any guest with at least one visit in the trailing 365 days. Guests with zero visits in 365 days are moved to a separate "Churned" segment and excluded from tier calculations.

Score processes data under a data processing agreement (DPA) with each restaurant operator. All PII is pseudonymized at ingestion — Score never stores full names or contact information; guest identities are represented by a deterministic hash of (email_normalized + phone_normalized).

Score is SOC 2 Type II compliant. Data is encrypted at rest (AES-256) and in transit (TLS 1.3). Restaurant operators retain full data ownership. Score does not cross-sell or share guest data between unaffiliated operators.