Domino's Pizza
Operational AI across the order chain: preparation-time forecasting and a genAI management assistant
Domino's raised its order forecasting accuracy from 75% to 95% with machine learning, and entered a five-year AI alliance with Microsoft (Azure OpenAI) covering more than 20,000 stores in more than 90 markets.
Key points
- Operational AI: order-time forecasting and a genAI store management assistant.
- In-house forecasting model plus an assistant on Microsoft Azure OpenAI Service.
- Forecasting accuracy from 75% to 95%, training cut from 16 hours to under an hour.
- Evidence B, confirmed status.
Objective
Make the order-to-delivery chain more reliable: predict when each order will be ready to staff better and inform the customer, and offload managers from admin tasks so they can focus on service.
The deployment
Domino's uses AI on two operational fronts. A forecasting model, trained on five million orders, estimates when an order will be ready; it raised the accuracy of those forecasts from 75% to 95% and cut the model's training time from more than 16 hours to under an hour. In October 2023, Domino's sealed a five-year alliance with Microsoft to build, on Azure OpenAI Service, a genAI assistant helping store managers with inventory, ingredient ordering, and scheduling, and to simplify and personalize the ordering journey. The group operates more than 20,000 stores across more than 90 markets.
Results Proof B
Quantified forecasting result (75% to 95%) documented by established tech press, and an AI alliance formalized by an official Domino's/Microsoft press release. The genAI assistant was in a pilot phase at the alliance launch, hence stronger scale evidence on forecasting than on the assistant.
How it works
Documented architectureThe stack in detail
- plateforme Azure OpenAI Service Cloud LLM service on which the genAI store management assistant (inventory, ingredient ordering, scheduling) and the simplification of the ordering journey are built.
- integrateur Microsoft Technology partner in the five-year alliance announced in October 2023; co-builds the genAI pilots with the Domino's teams.
- outil Modele de prevision de commandes (in-house) Domino's proprietary ML trained on five million orders; accuracy up from 75% to 95%, training time cut from more than 16 hours to under an hour.
- infra Application et suivi de commande Domino's Channel that consumes the preparation-time forecasts to inform the customer and set staffing.
How it runs, concretely
For ops teams-
1Order ingestion Data team
Live orders and history feed the forecasting model.
-
2Preparation-time forecasting AI
The model estimates when each order will be ready to set staffing and inform the customer.
-
3Management assistance AI
The genAI assistant helps the manager with inventory, ingredient ordering, and scheduling.
-
4Manager decision Store manager
The manager validates or adjusts the operational recommendations.
The historical and real-time order flow. Without clean, up-to-date order data, preparation-time forecasting degrades and staffing goes off.
How your customers perceive this type of use
Sourced studiesLe pricing algorithmique est le terrain le plus inflammable : 68% des consommateurs disent se sentir leses quand les marques utilisent le pricing dynamique et 80% jugent plus dignes de confiance les marques aux prix constants (Gartner, 2024). L'equite percue varie selon le secteur : le pricing dynamique n'est juge juste que par 33% a 40% des repondants selon qu'il s'agit de concerts ou de cinemas (YouGov, 17 marches). Le prix personnalise par les donnees individuelles est le plus rejete : 47% des Americains s'y opposent fermement (Consumer Reports, 2024).
Acceptance conditions
- La constance des prix comme signal de confiance : 80% jugent plus fiables les marques aux prix stables (Gartner 2024)
- Le secteur conditionne l'equite percue : le pricing dynamique est mieux tolere pour les cinemas (40% le jugent juste) que pour les concerts (33%) (YouGov 2024)
Red lines
- Le pricing dynamique percu comme abus : 68% se sentent leses (Gartner 2024)
- Le prix individualise a partir des donnees personnelles : 47% d'opposition ferme (Consumer Reports 2024)
- Les frais caches et hausses imprevues, vecus par 79% des consommateurs sur un an et associes a la perte de confiance (Gartner 2024)
Sources: Gartner 2024 · YouGov 2024 · Consumer Reports 2024
How to replicate
Inference, not sourcedData prerequisites
- a large, clean order history
- a real-time order flow
- inventory and scheduling data for the management assistant
Org prerequisites
- a data team or a cloud partner
- managers trained on the assistant
- integration with the ordering and tracking system
Possible stack
- a forecasting model (ML)
- a cloud with an LLM service (Azure OpenAI, GCP, AWS)
- an ordering app with tracking
The plan, step by step
- Step 1Consolidate the order history (volumes, timestamps, preparation times) and make the real-time flow reliable.Deliverable: Clean training dataset and a data pipeline in place
- Step 2Train a first preparation-time forecasting model and backtest it against history.Deliverable: Model validated offline with a measured baseline accuracy
- Step 3Integrate the forecast into the ordering app and staffing across a set of pilot stores.Deliverable: Pilot in production on a few stores with accuracy tracking
- Step 4Generalize the forecast across the network and track accuracy continuously.Deliverable: Full rollout and an accuracy dashboard
- Step 5Launch the genAI management assistant pilot (inventory, scheduling) on the chosen cloud LLM service, with a group of managers.Deliverable: Assistant in pilot with manager feedback collected
First step: Consolidate a usable order history and train a first preparation-time forecasting model.
The same pattern, proven in other industries
Sources
- S1 Domino's and Microsoft Cook Up AI-Driven Innovation Alliance for Smarter Pizza Orders and Seamless Operations Primary archive pending
- S2 How AI helped Domino's improve pizza delivery Secondary archive pending
- S3 Domino's and Microsoft to Create AI-Focused 'Innovation Lab' Established press archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.