Projecting demand for a signature product that must always be Hot-N-Ready

Production Management 3.0

Imagine you are a business who is well-known for a food product that is always hot and ready. When a customer shows up at the storefront in search of a delicious pizza pie, they expect a product that lives up to its name. Little Caesars wanted to live up to its signature claim of serving Hot-N-Ready pizzas, without creating waste or having too few pizzas. In order to determine how many pizzas would be necessary for a given location at any given time, they needed to understand what the key factors were that influenced demand for pizzas? Working closely with Little Caesars, the Microsoft team created the idea for the Perfect Pizza Kitchen. The team implemented the pilot in Little Caesars’ Test Kitchen a few months prior to OST getting involved in the project.

The primary goal was to demonstrate that the machine learning engine could deliver predictions to the store so that the system that was working in the store would receive projections. The system would then be able to determine, for example, you need to make this many of these pizzas at this time.

This project encountered challenges that the team worked together to overcome along the way. To begin with, we worked with Little Caesars to determine how the system uses data to make projections. We used two factors to determine the number of Hot-N-Ready pizzas required at each location.

The general thought was: if you know where the store is located, you can use the Weather Service data and the weather forecast as one of the indicators. Weather presumably has some influence on pizza sales.

The other key indicator was the schedule of major sports teams in the area. It’s safe to assume that sometime before, during or after a Detroit Lions game, for example, there will be predictable variance in the pizza sales of that area.

Using those two pieces of information, along with historical sales looking at previous sales data, the system combines those and makes a prediction.

We created a system that leverages machine learning algorithms to give stores hourly projections on Hot-N-Ready sales to make a prediction for a specific store and the number of Hot-N-Ready pizzas they will need to meet demand on a given day. Our team created an inference engine. What that means is, data without an answer goes into the system. The machine learning process takes messages and determines, ‘This is what the forecast for pizza creation has been, what are my projections going to be?’ Data with an answer gets sent back to the stores. The typical machine learning system has one model that feeds in data and comes out with a projection. This cutting-edge technology provides a tailored answer for each specific store.

Over time, the machine learning algorithm learns from its own level of accuracy. It self-corrects over time. There is still care and feeding that has to happen before the machine’s conclusions will be productive for Little Caesars.

The Results:

The biggest success is seeing everything work together in harmony. The project included having the system send sales data to the cloud, working on data, making sure it is managed and taking historical data and then sending messages downstream. Our team made sure the mechanics were working together and were fluid. That was the biggest success of the project. This project was also successful by being the first time this technology has been used at this scale. Our team also overcame hurdles that we ran into. This was a successful partnership of multiple departments – including a delivery lead, Data Analytics team and our principals. The technical solution speaks to the partnership and level of commitment between all parties involved. We have passed along the technology to the Little Caesars team. They have the technical capabilities in hand and are preparing for launch.

The outcome? We helped Little Caesar’s deliver on their signature promise of providing Hot-N-Ready pizzas. Our technology helped them achieve their promise. In the future, the hope is the technology will be deployed in multiple Little Caesar’s locations to provide accurate predictions for Hot-N-Ready pizza sales.

Facing Business and Technology Challenges?
Let’s Talk

Clients come to OST because we have highly capable teams who can design and build exceptional mobile applications, cloud infrastructure, custom software and more. Clients keep coming to OST because we bring decades of business and consulting experience into every project we undertake. If you want help solving for the complex challenges your business is facing, we’d love to start a conversation.

We look forward to hearing from you!