Understanding Modelling Different Pricing Effects

In the last few segments, you saw the different kinds of statistical models that you can build for market mix modelling (MMM). However, you did that mainly focusing on the advertising KPIs. Now, you will learn about the KPIs that you can build keeping pricing and promotions in focus.

Same as advertising, you observe the following typical patterns of price response:

• The current effect of pricing is the change in sales, revenue or traffic by the time the price has changed. Contrary to advertising, much of the pricing effect is immediate and, therefore, the current effect can be measured easily.

• There are two reasons for the carry-over effect of pricing: (a) consumers have current inventory stock, which is why they take time to respond to the price change, especially if they are perishable products, and (b) consumers may take time to realise the price change as they don’t always remember the right price of the goods they purchase. One special type of carry-over effect is a 'sale cut' post a promotion, as customers stock non-perishable products and hold back on the regular purchase.

• The competition effect of pricing is the change in sales, revenue or traffic due to a price change in competitive stores, especially for products that are common among them, or those that have a sticky price impression in the customers' minds.

Promotion Price Effect

Generally, you will have weekly information on the MRP, list price and discounted price for various items that you can use in the market mix model.

Calculate the shelf price inflation using the list price or base price. This will show how the base price has changed week over week.

Calculate the percentage discount offered using (a) the list price and discounted price, and (b) the MRP and discounted price. This is to check human behaviour such as whether people do the mental math against the MRP or the earlier list price. Either way, you will get the effect. Since these two KPIs will be highly correlated, the model will ultimately choose only the one that will be of the highest significance and the best explanatory factor.

The coefficient of the list price effect will be negative and that of the discounted price effect will be positive. This is because if the list price goes up or if the shelf price inflation is too sharp, it should result in declining sales, whereas with an increase in discount, sales should increase.

Reference Price Effect

This is similar to shelf price inflation, with the difference that instead of comparing with the previous day’s or the previous week’s price, you compare with an average price of the previous 2/3/4/6 weeks. Thus, shelf price inflation is a specific form of the reference price.

Reference price could also be calculated with respect to the competition price if you have competitor price information. Thus, if you know the competitor’s price for the same SKU, then you can calculate how pricey the products in the target store are compared with your competitor store.

Interaction Effect

One of the biggest aims in MMM is to model the interaction effects. This is generally introduced in a multiplicative mode in MMM. For example, if you believe that advertising and discounting have a stronger effect on sales, then you can introduce a new independent KPI that is a product of both, along with their individual KPIs. The coefficients of individual KPIs will show the effect of the singleton variable, whereas the co-efficient of the multiplicative term will show their joint impact.

Comprehension

The weekly list prices of two competitive products from different brands are given in the table below: