November 23, 2012
By Pelin Thorogood,

The New Science of Retaining Customers

Consider this familiar story. After promising initial purchases, your marketing team observes a steep falloff of buyer interest for some of your new products. In contrast, other offerings are doing surprisingly well. These offerings not only attract repeat purchases, but also more frequent ones among key buyer groups. Unfortunately, it isn't clear what has caused your best customers with the highest revenue potential to buy, leaving you without the information you need to repeat your successes or avoid the pitfalls. Where do you start to plan your holiday campaign? The mountains of data you collect were supposed to show you the way. Instead, they are stuck in disparate silos, inconsistent across marketing channels and out of date by the time reports hit your desk.

Mining the insight from that data can be as exasperating as trying to solve a three-dimensional puzzle. Think of it as a Rubik's Cube, with the sides of the smaller cubes representing all the variables impacting customer retention—the campaigns, offers, and advertising across all marketing channels—and the data generated by each. The combinations seem infinite.

Customer retention and, ultimately, customer lifetime value (CLV) are the end goals of CRM. But how do you evaluate which customer retention and upsell efforts work best with your buyers? How did customer satisfaction scores, order history, and engagement frequency correlate to purchase and repeat purchase behavior? What marketing campaigns were most effective, and with which buyer segments? And, across the holistic customer journey, what activities supported loyalty or led to churn?

Real-Time Insights into Buyer Behaviors

When it comes to metrics, we normally focus on the top-line numbers. We track customer satisfaction scores, but they don't necessarily reveal much about a buyer's interest in repeat purchases or customer retention over time. We identify revenue per customer, but without analyzing campaign effectiveness, by user segment, across channels to identify the most valuable buyers. What's needed are metrics delivering actionable insight into the actual levers driving customer retention. Only by understanding buyer behaviors at a deeper level can you develop effective, new selling strategies and the ability to target campaigns and personalize offers to drive that all important CLV.

Fortunately, the rapidly emerging world of digital analytics enables marketers to gain insight into customer-buying behaviors in ways that were impossible just a few years ago. Digital analytics—the ability to collect, analyze, and visualize information from big data in real time—is the key that unlocks an extraordinary amount of information about your customers: how, when, and why they buy; what causes churn; and the best ways to retain them.

There are two transformational principles in the world of marketing and analytics:

1) Multichannel Marketing: With the explosion in social and digital media, your buyers engage with your brand through an increasing number of channels, making the customer journey more complex every year. This is the era of multichannel marketing, with the mix now including an array of traditional and online channels, combined with the world of social and mobile apps.

2) Paid, Owned, and Earned (POE) Interrelationships: These marketing channels and their performance are tightly interrelated. Siloed information about them lacks context and is of little value. More than simply evaluating individual campaign results, the marketer needs to understand the interrelationships among paid, owned, and earned (POE) media. Social buzz from a viral video, for example, may amplify your ROI in online banner ads, measured by both traffic and buyer conversion rates. It is only by analyzing correlations between competing factors that you understand individual campaign and channel performance trends over time. This is essential to producing a longer-term picture of factors contributing to customer retention.

Complexity and Surprises

Let's look at these principles through the lens of a use case involving an e-commerce business with tens of thousands of member buyers worldwide. These members interact frequently with the company and each other, and they have the opportunity to take advantage of numerous special offers online and through call centers. The goal set by management: increase per-buyer revenue across all product categories by analyzing purchase activity by customer segment in all channels and campaigns.

The company has been able to measure customer churn, but is much less clear about the factors leading up to it. What campaigns and offers drive repeated sales of multiple products, and how does buyer reaction differ for different segments? How does purchase recency and frequency relate to the likelihood of churn or new purchases? Digital analytics makes it possible to correlate key demographic, behavioral, social, and financial factors to decipher obscure relationships and dispel misconceptions.

In this case, the marketing team benefits from analyzing marketing channel and campaign effectiveness against factors such as:

Number of orders
Average order size
Order recency
Order frequency
Number of items ordered

The analytics process produces surprises. The team observes unexpected correlations and discovers none where they were projected. Now they have the actionable data they need to increase the likelihood of certain behaviors that produce more purchases or reduce customer churn.

Viewed holistically, we see many stakeholders benefiting. Product category managers assess product viability correlated with marketing spend and loyalty programs. The marketing team learns how campaigns and channels stack up in terms of churn and loyalty, as well as return on marketing investment (ROMI). Finally, finance benefits from analysis relating discounts, customer loyalty, and effect on margins.

Customer Retention—The New Science

All this tells us that the intelligence needed to drive customer loyalty must both accurately describe a business's current state and create a foundation for forecasting future behavior. Calculating CLV is an exercise in predictive, multichannel analytics. To achieve this, your analytics capabilities need to connect the dots among your many data sources, provide real-time analysis, and predict relationships and trends that ultimately enable you to discover new truths.

The good news is that all of this is possible today. Indeed, think of this as the new science of retaining customers, drawing on fresh methodologies and a new generation of technologies in multichannel digital analytics.

Pelin Thorogood is the chief marketing officer of Anametrix and a specialist in new media marketing and analytics. She was formerly the CMO of WebSideStory (acquired by Omniture/Adobe), the first cloud-based Web analytics provider.