Article: Tuesday, 14 July 2020
Companies spend significant amounts of money, time and effort to acquire new customers. They can't afford to lose these hard-won customers. But keeping customers happy, engaged, and of course, loyal, sometimes means offering financial incentives such as discounts or gifts to stop them defecting to other suppliers. The tricky part is figuring out exactly which customers should be targeted. Recent research by Dr Aurélie Lemmens, Associate Professor of Marketing at Rotterdam School of Management, Erasmus University (RSM) and Sunil Gupta, Professor of Business Administration at Harvard Business School, addresses this question. They offer a new algorithm that firms can use to maximise the profits of retention campaigns. Lemmens and Gupta argue that companies often target the wrong customers. They propose a new method to target those customers that create the biggest boost to a company's profits.
Customer attrition (also known as ‘churn’) commonly affects many organisations and companies in a variety of industries. For example, most North American credit card providers deal with annual churn rates of about 20 per cent, European mobile phone carriers deal with 20 to 38 per cent churn. Charities face high churn rates too.
To tackle churn, companies develop sophisticated churn management strategies that involve two steps: ranking customers based on their churn risk, and then offering incentives to a core group of customers who appear at the top of the churn ranking. Retention budgets have consistently eaten up a large share of companies’ marketing budgets, and created a need for methods that make retention actions more cost-effective.
Dr Aurélie Lemmens says these traditional methods of trying to retain customers are flawed. The two steps are not set up to maximise bottom-line profit of the retention campaign because they ignore a simple but important fact: not all customers are equally important to the firm. She further explains: “They focus on the customer's churn risk, but not on the incremental profit the firm can make by targeting particular customers. Some customers are likely to leave but there is nothing the firm can do to retain them. In fact, even among the customers that can be retained, some are worth larger investments than others because they bring in more value to the firm than others.
“In contrast to prior approaches to churn management, our approach focuses on the return-on-investment – which we call profit lift – of the retention interventions. We demonstrate the benefits of using an approach that estimates the financial impact of a targeted marketing intervention. Our findings highlight the need for marketing academics and practitioners to pay attention to the choice of their objective (or loss) function, a feature that is often neglected in model estimation processes. This choice should match managers’ objectives.”
For companies to have insight in this overall profitability, they should consider how much more customers would spend, and much longer they’ll remain as customers, if they get an incentive – a retention offer, compared to a situation where they would not get one. The researchers’ predictive model takes all these factors into account, and determines the optimal number of customers to target. Of course, companies don't want to lose customers, but targeting too many of them can be too expensive to be worthwhile.
The study demonstrates that optimising which customers and how many of them to target has a substantial impact on firms’ profits. Two field experiments found this approach to be significantly more profitable than the ones that firms currently use. In terms of impact, the authors speculate that a single retention campaign using their method could increase a firm’s overall profits by at least 4 per cent.
For firms, it is important to know that these extra profits come with few implementation costs. Lemmens and Gupta explain: "All we did was use the same incentive you would otherwise use. There was no change in anything except who you target, so there was no additional cost."
In general, this research provides broader implications beyond churn prevention for academics and firms. The approach fits many contexts, within and outside marketing, in fact wherever organisations seek to target a set of individuals with a specific intervention (e.g., catalogue mailing, charitable giving, and personalised medicine).
The key message to remember is that, when building their own ‘goal-oriented’ objective functions, decision makers should
Ensure that they predict the true outcome of interest (i.e., goal of the intervention).
Use a weighting scheme that prioritises customers, donors or patients that have the largest impact on the success of the intervention.
This is relevant even in non-profit contexts such as predicting patient compliance with medical treatments. In this case, the objective function could incorporate patient-specific health risks and benefits associated with complying with the medical treatment.
The algorithm for this new method is called ProfitBoost, and it uses the R platform for statistical computing and graphics. It will be available as a package soon: see www.aurelielemmens.com for the latest updates, or email firstname.lastname@example.org for more information. Right now, a Dutch telecoms company is also investigating how to implement this model in its daily operations.
RSM offers a variety of courses and programmes about artificial intelligence, suitable for executives and managers from all disciplines. Find out more about data analytics in our open education courses.
Rotterdam School of Management (RSM)
Harvard Business School
RSM offers Executive Education and Master programmes in various business areas for any stage of your career. For instance:
Corporate Communications & PR Manager
Science Communication and Media Officer