How businesses can use data analytics to predict (and control) their demand
How much fuel a single petrol station sells in any given week is determined by a myriad of variables.
There’s the petrol station’s prices, the proximity of its competitors and their prices, the overall market price, and the timing of holidays and long weekends.
If we can understand how these variables work together we can then accurately forecast future sales for petrol retailers and indeed for any other business, allowing them to optimise their operations, with the ideal number of staff and the right amount of stock and equipment.
Drawing on a technique known as time series modelling, we go back and look at past sales and identify patterns so we can determine how those variables affect demand. We can use that information to predict demand in the future and reveal the levers a business needs to pull to get the desired outcome.
The businesses can effectively control their demand.
Measuring effectiveness of discounting and promotions
For instance, in the case of the petrol station, analysis might reveal that if they dropped their fuel price by 2 cents per litre they might sell an extra 10,000 litres a week, more than making up for the lower margin. Alternatively, analysis might reveal that 2 cents per litre price rise will reduce sales volume, but with the higher sales margin the fuel retailer will earn more.
Petrol is a very price elastic product and motorists will deliberately go out of their way for cheaper fuel, particularly now with record high petrol prices. People don’t enjoy buying petrol, however it is a necessary purchase to run their car, so consumers are often very price sensitive.
Demand forecasting isn’t just confined to sales. It can be used for resourcing optimisation, for example a government department running court services could forecast the future number and type of court cases and then determine the number of judges, support staff and courtrooms that will be needed in the future.
It could also be used for a consumer products business to determine the effectiveness of promotions and forecast volume based on a given promotion calendar. This would involve drilling into the past four years of sales data across dozens of different product lines to see when promotions were running on some products, their effect, how much the higher sales cannibalized their other products, and how much market share they took from competitors. The final result could provide the business with accurate and detailed information about the effectiveness of their different types of promotions and help them design better promotions in future.
Eliminating food manufacturing wastage.
Demand forecasting is especially important for manufacturers of perishable foods such as dairy products. They don’t want to produce too much and have product spoil, but on the other hand they don’t want to produce too little and miss out on sales that they won’t get back the following week. Demand forecasting allows them to predict the right amounts of raw materials to produce enough product to walk the line between too much and too little.
Alternatively, they can pull the different price and promotions levers to adjust demand to meet their output.
With the right data analytics, demand optimisation can be a powerful tool for business.