How to Use Data to Drive Customer Propensity

Scott Logie

December 11, 2019

Propensity – that is, a natural inclination or tendency – is a term that’s used frequently in Marketing, but not much anywhere else. Customers are a jumpy lot: a recent survey from the DMA found that 61% of customers had switched at least one brand in the last 12 months, so it’s no wonder that marketers and brands want to know what their customers will do next and when. Do they have a propensity to purchase or donate again or to lapse or go elsewhere? To try and predict these answers, propensity models attempt to anticipate customer behavior and trends. So how do we go about figuring out what customers are likely to do next and finding the people who are more likely to do the thing we want (or don’t want) them to do?

It comes down to probability and the likelihood of something – a certain behavior or activity, for example – happening. And that likelihood can allow you to predict which other existing or potential customers will also do the same thing if you know a little bit about them. In this scenario, certain predictions can be made, as long as we have the right data to understand the factors driving the decision.

How to Improve Propensity

There are two ways to look at this: firstly, at an individual level (e.g. ‘What is the likelihood that I will buy this product today?’) and secondly, from a group perspective (‘Of all of the people we are interested in, who else will buy this product today?’). If you can crack both of these, you can hopefully sell more products or drive the behavior you’re trying to achieve.

You can apply this same way of thinking to many customer activities: purchasing a product, taking a holiday, making a donation or renewing an insurance product. The key is to find the people with the propensity to do each of these things.

The Next Best Action

At an individual level, what’s known as the ‘next best action’ can help a brand better predict what each person will do next based on what they and others have done in the past, by looking across all of the different options and picking the one they believe is most likely. This can be done more accurately by taking into account data such as past behavior, current status (what they last bought, when and how much they paid) and the behavior of similar customers. This should then be combined with the product being sold to them, to ensure the transaction is a positive one.

By balancing all of this data, it’s possible to create a model for each person that scores all the likely events and picks the one most likely to happen. If necessary, this can be weighted towards where more profit can be made, and this can then be suggested to the customer at the appropriate time to help inform their decision.

At the group level, the application is similar but with a slightly different outcome. Here it is necessary to apply a score to everyone and pick those with the highest likelihood, probability or propensity to do that thing. For example, when looking for people to take part in a particular event, create a model that scores everyone and then pick those at the top who are more likely to do that event.

Data Holds the Key

It’s clear to see that data underpins both of these approaches: specifically, the right data and enough data to allow propensity models to be built and applied. Generally, that data exists inside a business: it’s the data you have on existing customers as well as past behavior and outcomes (e.g. who lapsed when and why). External data can also contribute by helping to formulate a different view of a consumer, such as what they do outside of the business, online, on social media and beyond.

Harnessing this data and applying it to customers and the business’ objectives can help improve the Marketing department’s ability to predict customer behavior, as well as improving their own propensity to make better decisions. The more data, the better informed the decision-making, the better the bottom line.