Variability in number of customers can be very costly

A restaurant’s food supplies can be insufficient if their number of customers is high or go to waste if it’s low.

A department store can be unable to serve a high number of customers if understaffed or spend too much on staff costs if the number of customers is low.

To get a feeling of the size of the problem, consider that overstocks and out-of-stocks are responsible for 7.3% in lost revenue for the average retailer, for a total $1.1 trillion globally.

Why do companies need accurate forecasts?

The real problem is not the variability by itself but the discrepancy between the number of customers and the predicted number of customers that was considered at planning time.

Every company makes these forecasts, but what matters is how accurate the forecasts are. The more accurate the forecast we use to plan is, the smaller the lost revenue will be.

With a good forecast our restaurant will order the right amount of food supplies, our retailer will keep its customers happy while avoiding redundant inventory and the staff scheduling of our department store will be correct.

CASE STUDY

Forecasting Gym Attendance to Optimise Staff Scheduling

ABC-Gym offers a low-cost base membership and many add-on walk-in classes.

ABC-Gym’s revenue comes mainly from the classes, which the customers love because of the close interaction they get with the instructors. In fact, some of these classes only have 3-5 spots to ensure high quality workouts.

ABC-Gym works with many freelance instructors paid by the hour and sends them their rota one week in advance.

The problem

The challenge that ABC-Gym faces is that when they plan the instructors’ rota (supply) they don’t know exactly how many members will show up at the gym the next week (demand).  

Additionally, gym attendance varies significantly during the day and throughout the year. 

The solution

When ABC-Gym came to us, they were using an empirical forecasting approach: they would use this week’s attendance levels as a forecast for next week’s. Although simple, this approach tries to address intraday patterns and seasonalities. 

Unfortunately, on many days the actual number of customers attending the gym was considerably off the forecast with the following result:

  • Actual > Forecast: shortage of classes. Unhappy customers and lost revenue
  • Actual < Forecast: unfilled classes. High instructors cost for low revenue

In just 4 weeks, ⍺I Demand Forecasting allowed ABC-Gym to use all their gym usage and customer data and augmented it with day-specific information (weather forecast, calendar events) to obtain better predictions.

Business results

With ⍺I Demand Forecasting, ABC-Gym achieved a 40% improvement in forecasting accuracy, which translated in:

  • 10% increase in the profits originated from classes and
  • Increased customer satisfaction due to better class availability

Project Highlights

  • Project delivered in 4 weeks
  • 16 months of data analysed
  • Results: 40% improvement in forecasting accuracy, which translated in 10% increase in classes profits

⍺I Demand Forecasting

Your AI support platform that helps you improve planning and inventory optimization by analyzing all your relevant data and learning the unique patterns of your metrics.

Product Trials

We provides 6-12 weeks trials to tackle your company’s operational challenges and demonstrate the business value that Alpha-i can add to your organisation.