As a consumer, if you’ve ever tried to cancel a product or service, you’ll likely have encountered a reactive customer retention strategy. This is when your newly out-of-favour brand does everything it can to persuade you not to leave. It’s a last-ditch effort.

And brands don’t like to let you go without a fight. We’ve all been there. We’ve finally made the decision to switch broadband, or that we don’t really need that fancy box of fresh food delivered to our house every week. We look on the website. Turns out we have to call a number to cancel. That’s odd, we think, as when we signed up we did it on the website in under 20 seconds. Oh well, we call the number and sit through the options from the automated assistant. We press for the ‘thinking of leaving us’ team. And then we’re met by the ever-so-courteous (read highly trained) cancellation assistant as they ‘listen’ and walk us through their script. We eventually get to the part where they offer us discounts, upgrades and ‘goodwill gestures’. They keep us on the phone forever, absorbing our frustration minute by minute, until we eventually decide whether to stick or twist.

Clearly, this is a critical moment in the customer journey for brands to get right. The benefits of retaining customers are well-documented. From a marketing perspective, targeting is a non-issue. Customers have to call, and even helpfully segment themselves into those that are looking to cancel, so they can be automatically directed to a well-staffed, highly trained team in the customer service centre. But this approach also has obvious drawbacks. Here are just a few of them:

• You won’t save everyone
• It’s very expensive (both through staffing call centres and giving discounts)
• These customers may already have complained to everyone they know about your brand

The reactive model is the last line of defence a brand has, and so it’s critical to make a good job of it. Analytics can help improve reactive approaches to customer retention, typically by identifying the best offer to save the customer. Such an offer can be highly personalised and targeted, generated by analytical decisioning models, and made available to the front-line customer service teams to be used within the ‘save’ conversations with customers. Analytics can also be used to identify customers that you do not want to save due to their lack of profitability (either historic, current or expected future profitability) and again this information can be provided to the front line.

But when it comes to the important business of retaining customers, a reactive approach shouldn’t be the only thing you do.

Proactive customer retention

A proactive customer retention model is one tool that organisations use to prevent customers ever picking up the phone to cancel. Using machine learning techniques, these models are trained to identify signals in your customer behaviour data that indicate customers are likely to attrite, often long before they decide to cancel. They are used to identify the customers at most risk of leaving, allowing savvy marketing teams to be proactive in their contact strategy, and less reliant on the expensive reactive approach.

In some ways it’s harder to do, but with strong analytics and a bit of focused effort, brands can nip many of those cancellation thoughts in the bud. Doing it right will see customer retention and customer satisfaction figures improve and, due to the unexpected nature of it, can turn dissatisfied customers into brand advocates. Not to mention the cost benefits of a reduction in customer service calls.

For example, if you’re working for a food delivery company, you might learn that customers are more likely to cancel after they’ve had a service issue within the first three weeks of signing up. In this case a smart move might be to build a recurring campaign specifically for those segments, acknowledging the issue and giving them a discount for the next month. The insight may also help you identify and solve the service issue itself and prevent any future cancellation thoughts before they come up.

Perhaps you work for a bank and your analysis highlights that early warning signals of attrition include a reduction in debit card usage or fewer logins to the mobile app. Again, a good move here might be to address the behaviour through a targeted campaign that seeks to rebuild the relationship with these customers.

Or maybe you work for a supermarket retailer and the behaviour you want to focus on is weekly shoppers that suddenly become less frequent shoppers.

Depending on your sector, product and audience, the things that indicate customers are likely to attrite will vary considerably. What is consistent, however, is that you can often use existing behavioural data to predict likelihood of attrition in future, and use that insight to help you develop tailored marketing activities that address minor issues before they become big problems.

Isn’t proactive retention just good marketing?

No, although their goals are very similar, proactive retention models aren’t to be confused with general marketing. A good proactive retention strategy is one that specifically acts on flags in the data that analysts have shown to be significant drivers of customer attrition. As with any statistical model, accurately predicting which customers are likely to leave is never going to be correct 100% of the time. They also need to be combined with models that tell you which customers you will be able to persuade to stay through your marketing contacts – uplift models.  

Clearly developing proactive retention strategies demands good data to work with, an analytics team that knows what to look for, and a marketing team that’s capable of taking appropriate action. And we should also point out we’re not advocating adopting only a reactive or a proactive customer retention approach. The most successful brands will have both, each optimised, and working in conjunction. Strong analytics is an indispensable tool in both proactive and reactive strategies, essential for deciding where to focus, identifying how best to retain customers and measuring the success of new initiatives.