According to Nielsen, the majority (60 percent) of supermarket sales in New Zealand are on promotion. But do you know whether your buy one, get one free promotions are more effective than your half-offs? What promotional structure drives trial of a new product and whether or not that will grow the category?
As the old adage goes, you can’t manage what you can’t measure. Therefore, to improve promotional performance we must first measure it. Point-of-sale data, or better yet, loyalty card data, can be analysed to measure performance of a promotion on two key dimensions: whether it delivered financially for the retailer and whether it was valued by the customer. It sounds straightforward, but the first of these is often mismeasured and the second is ignored entirely more often than not by retailers around the world.
To understand the first issue, you need to understand a bit about how promotions can be measured. As pharmaceutical testing has done for years, the only way to accurately measure the impact of a promotion, or any piece of marketing activity for that matter, is A/B testing. This is where one randomly selected group is exposed to the promotion and another randomly selected group (the control) is not. The mistake often made in measuring the financial impact is that people can underestimate the impact of a control on the measurement. The trouble with this method is that it’s not always possible to randomly expose only one group to it. The trick of the method is using analytical sleight of hand to best estimate the behaviour of this control group when a perfect control is not possible. The quality of this estimation is then what determines the quality of the measurement. Poor estimation will lead to measures of financial performance that are totally unrelated to the true performance of the promotion.
A more troubling aspect is that whether or not the customer values the promotion is usually not even considered. It’s easy to think that if the promotion is delivering a financial return then it doesn’t matter whether customers value it or not, but this is a short-sighted view. From our work here at TRA, we know how important promotions are to customers’ purchasing decisions. According to a Stuff.co.nz poll, only 24 percent of people claim to buy the products they want, regardless of the promotions. It’s clear that promotions have a huge impact on customers’ experience of a retailer. In today’s competitive environment where customers vote with their wallets, a poor experience will lead customers to switch to a competitor, while a great experience can lead to a loyal customer or even an advocate.
Measuring all promotions quickly leads to a bank of knowledge about what works and what doesn’t. This bank of knowledge is where the value in the method really lies. Leveraging it doesn’t only answer the questions at the beginning of this article, but it can also be used to predict the performance of a proposed promotion. Like many machine learning processes, the more data we feed, these models the better they get. The more types of promotions that we try, the greater the accuracy when predicting the promotion’s performance. Forward-thinking retailers are already deploying this approach with great success. For such an important part of the customer experience, why wouldn’t you?
This copy originally appeared in NZ Retail magazine issue 740 October / November 2015.