Public customer case study
A public case study for a 12-unit upscale dining operator using Before Service to generate Pre-Service Briefs from POS, inventory, supplier invoice, and labor signals.
Pacific Northwest, USA · 42,000 average monthly covers · March 2025-present
General Managers, Executive Chefs, and Beverage Directors manually aggregated Toast POS reports, MarketMan inventory snapshots, and 7shifts labor schedules in Excel during pre-shift hours. The baseline was 75 minutes per unit each day, with a 14% discrepancy rate between forecasted COGS and actual end-of-shift COGS.
Pre-Service Briefs combine POS historicals, live inventory counts, supplier invoice PDFs, and labor availability logs into one review draft before service.
Pre-Service Brief generation fell to under 4 minutes. The discrepancy rate between forecasted and actual end-of-shift COGS fell from 14% to 4.2%.
Measured through GM tablet time-tracking logs and reconciled weekly P&L reports from March 2025-present. Public disclosure and full name/title attribution are approved.
“The primary friction point was always the lag between the supplier invoice and the menu price. We were flying blind on margins for three to four days a week. Moving that visibility to a pre-service brief allows us to pivot specials or adjust portions before the first ticket hits the kitchen, rather than realizing we lost margin at the end of the month.”
Human approval boundary
Before Service takes no automatic action. Every pricing adjustment, labor cut, and inventory order must be reviewed and executed in the source system by a human operator.
Aggregate operating metrics
Measurement source: internal system telemetry and user-interaction logs.
Representative workflow patterns
Fast casual operator
ProblemLunch demand varied week to week. Pars were a guess and over-prep was difficult to trace.
ActionSwitched the team to a forecast-driven prep plan and re-batched sauces mid-service instead of pre-batching.
ResultThe intended outcome is a more reviewable prep plan and a clearer record of adjustments.
Evidence type · representative scenario
Permission · no customer identity used
Reporting window · illustrative, not measured
Café owner
ProblemCoffee margin looked stable while pastry ingredient costs changed after a supplier update.
ActionUsed menu margin analysis to flag the slip, re-quoted two SKUs, and adjusted one recipe gram count.
ResultThe intended outcome is an earlier operator decision on recipe, supplier, or menu pricing.
Evidence type · representative scenario
Permission · no customer identity used
Reporting window · illustrative, not measured
Multi-location manager
ProblemThree locations, three different prep notebooks, and no way to compare leak by site.
ActionRolled the morning brief out across all sites with a shared waste tag taxonomy.
ResultThe intended outcome is a comparable view of site-level risk before the weekly review.
Evidence type · representative scenario
Permission · no customer identity used
Reporting window · illustrative, not measured