In the first part of the series “Dynamic Pricing” you read that dynamic pricing describes the application of statistical learning to price optimization. A trader’s goal is to maximize performance with minimal price increases at constant costs. This is exactly where the Predictive Analytics software attacks, which can do much more than just observe the competition.
In order to find the optimal pricing model, it is important to weigh the available capacities against revenue maximization. On the one hand, price increases lead to an increase in revenues, on the other hand, the volume of sales decreases. For this reason, it is important to determine the price elasticity of demand. Price elasticity represents the real change in demand when the price varies. It can be derived from the price-sales function. The estimation of the price-sales-function is the core of the implementation of a target-oriented dynamic pricing strategy.
Determinants of price:
By forecasting influencing factors, it is possible to determine reliable estimates of price level and contribution margin in a future period. The price determinants in Dynamic Pricing are not only formed on the basis of external factors such as competitive prices, but also internal factors such as assortment, stock levels and the historical purchasing behaviour of customers, as well as their clicks or search behaviour. In addition, determinants (time of day, season, advertising campaigns) also flow into demand. Based on all these factors, price sensitivity is then measured using the price-sales function.
Requirements for Dynamic Pricing Analytics:
The self-learning engine has the task of forecasting the PAF extremely precisely from large amounts of data. To do this, the algorithms must constantly analyze the interplay between price and customer in real time. In the first step, the forecasting model is created on the basis of historical data. In the second step, the model is adapted to determine the individual price sensitivity on the basis of price reactions.
The aim is to derive a price-sales function for each individual article. The target variables define products and product groups, bestsellers or long-tail products, as well as core or marginal products.
A target-oriented, successful pricing requires a clear pricing strategy. Aspects such as increasing customer loyalty or acquiring new customers, but also product launches or sales must be integrated into a consistent strategic framework.
Let’s take a retailer as an example: XY sells wool sweaters. The price-sales-function of a wool sweater is estimated through permanent price tests. So-called price bands are set for this wool sweater: underprice, middle price and maximum price. We assume a wool sweater, which is currently offered for 40€ in stores and whose sales volume is about 5000 pieces.
From these data we form three homogeneous test groups for the wool sweater: In group 1 the current price is increased by 10% to 44€, in group 2 the price remains the same at 40€ and in group 3 the price is reduced by 10% to 36€. What for human processing quickly leads to a cost-intensive use of results can be evaluated with the corresponding models within a few moments. The test results serve as input for the Black Box. Thus significant influence factors for the purchase decision of the customer can be determined, and the optimal price be defined.
The Predictive Analytics Modeling developed by Dastani Consulting puts companies in the comfortable position of finding exactly the “optimal” price, where the customer has the maximum willingness to pay and the trader increases his performance. Dealers often cut themselves off from high profit potentials by selling high-quality products below the individual willingness to pay. The future promises high potential for using artificial intelligence in pricing.
Read the third part of our Dynamic Pricing series to find out how retailers can use Next Best Offer to further strengthen customer loyalty and better understand their customers’ spending habits.