Year
2023
Client
Christian Clay
Category
Hospitality & Food Service
Product Duration
3 - 4 Weeks
We spent two weeks working with the group's operations manager and head of finance to understand how decisions were currently being made. We found that the POS system contained rich sales and cover data that was almost never used analytically. Supplier invoice data was reconciled monthly rather than weekly, meaning cost changes were not reflected in margin calculations quickly enough to act on. We also ran a menu engineering analysis across all locations to establish a baseline profitability picture by dish.
We designed a weekly management digest that presented each menu item's margin, volume trend, and a recommended action: promote, hold, reprice, or retire. The digest was designed to be read in under 10 minutes and to lead with the highest-impact decisions first. We also designed an alert system that flagged any ingredient whose cost had risen more than 8% since the dish was last priced, prompting a repricing review before margin was lost.
We built a data pipeline that ingested daily POS exports and weekly supplier invoice data, normalised them into a unified profitability model, and applied trend analysis to each dish across a rolling 8-week window. A language model generated the plain-language recommendations and rationale for each item, which were formatted into an HTML digest and emailed to the management team every Monday. The ingredient cost alert ran as a separate daily check. Full build took 6 weeks including data normalisation, model calibration, and a 2-week parallel-run period before the team relied on it for decisions.
17% improvement in average dish margin across locations
