In a 2020 research of North American grocers, 70% of respondents said they could not consider all the relevant aspects of a promotion—such as price, promotion type, or in-store display— when projecting promotional operations. But they wish they could.
That’s the point with using predictive analytics for sales prediction. It is possible to understand the impact of each factor on the total sales.
But, for most businesses getting an exact sales forecast is nevertheless a significant challenge. Thanks to incorrect forecasting methods based on hunches, companies end up getting little understanding of projected sales. According to Clari, a revenue services platform, 93 percent of sales managers cannot anticipate revenue within 5 percent, even at the end of the quarter.
We have gathered sales historical data from various retail stores for several days. The data collected includes 20 criteria for each record, for instance:
There are 6569 records in the dataset used.
TADA machine learning’s approach is an excellent forecasting method. It can provide accurate sales predictions and help understand the factors influencing sales.
It poses the following sales and marketing question: Can daily sales be accurately predicted with machine learning?
Sales forecasting is a crucial ingredient of every expanding retail business. Without precise forecast-based demand estimation processes in place, it becomes almost unmanageable to have the correct quantity of stock on hand at any given moment.
Too many products in the warehouse mean more money strapped in inventory, and not enough could result in out-of-stocks — and drive customers to use solutions from your competitors.
TADA has selected the following four main criteria out of the twenty available in the dataset:
Not only does the sales manager obtain an accurate forecast, but TADA makes it possible to understand the rationale behind this sales forecast. And here, TADA is full of common sense. The number of customers entering the store is by far the most important influencing factor.
In other words, to increase sales and revenues, the number one objective for sales teams is to get more people in the store.
Another influencing factor in the predictive analytics tool is the distance to the closest competitor, with a weight of 13%. It is not the most critical factor. But it is interesting to note.
For instance, the dataset comprises the opening year of the competitors’ shops. And it does not matter in the prediction made whether the competitor has been around one year or five. Using the ‘live predict’ feature from TADA, we can analyze this criterion one step further.
We can see that competitors’ stores impact sales up to a distance of 2635 meters. Beyond this threshold, a competing store has little effect on sales predictions. Too far to count?
Improved predictability has two direct consequences. It prevents the merchandising team from launching bountiful promotions that would never give an ROI, thus avoiding expensive mistakes; two, it enables more informed negotiations on ordering inventory, preventing over-or under-ordering.
In one week, sales teams gained significant help in achieving their total sales by: