Making Big Data Actionable for Multichannel Retailers

Go to any retail marketing conference and you’ll probably hear some variation of the following two statements:

• "We don’t have enough data to make good marketing decisions."
• "We’re drowning in a sea of too much data and don’t have a way to make it actionable."

Every retailer recognizes that profitable customer acquisition, retention and activation require data-driven methodologies. Yet, a recent study from COLLOQUY noted that 60 percent of U.S. retailers said they don't have data reliable enough for executing effective customer acquisition initiatives.

Despite possessing extensive customer data — including variables such as preferred product details, promotion channel, purchase channel, web navigation and clickstream data, path to purchase, and email engagement — the majority of marketers lack the actionable insights necessary to fuel effective prospecting or to optimize marketing to existing customers.

With all the talk about big data, how can multichannel retailers make the best use of information and leverage it for measurable results?

Make Big Data Even Bigger … and Smarter
This might seem counterintuitive at first, but supplemental data from sources beyond your own customer files can yield valuable insights. If you’re like most retailers, you rely heavily on RFM (recency, frequency and monetary) data to target and prioritize your marketing efforts. RFM data continues to be the most powerful indicator of customer behavior, but it also has limitations. These three variables represent only a narrow slice of a customer’s total purchasing profile, revealing what they spend with you, but not their other interests and priorities in life, or the many purchases they're making elsewhere.

RFM segmentation often ends up grouping together customers who are actually quite different from one another. For example, let’s say you have two customers who have spent between $25 and $49 with you multiple times in the last 12 months. They may look similar from a RFM perspective, both in terms of 12-month purchases and even looking back at cumulative lifetime spending. However, the reality may be that one customer has been purchasing primarily from your stores and/or website, while the other is spending an equal amount with two of your competitors. From a RFM perspective, they look identical, but they need different marketing treatments based on their total purchasing behavior.

To combat this knowledge deficit — and improve the effectiveness of customer and prospect marketing efforts — retailers should strongly consider the use of outside data sources, whose data typically falls into two main categories: demographic/psychographic characteristics and transactional characteristics.

By appending externally sourced demographic/psychographic data to customer files, retailers can build a more comprehensive understanding of their customers and segment them more effectively. There are several sources for licensing this type of supplemental data, as well as providers of social overlay data. These incorporations can add insight into customer sentiment for better refinement of customer personas and audience segments.

Supplemental transactional data that includes customers’ purchasing behavior outside of your organization can be even more powerful. That’s because the greatest predictor of future purchasing behavior is past purchasing behavior. Data cooperatives offer the opportunity to benefit from data provided by other organizations, including retail and service businesses, catalogers, nonprofits, publishers, etc. This allows all participants that contribute data to benefit from a detailed, 360-degree view of prospects’ and customers’ total transactional behavior. Predictive modeling that leverages this large and diverse data can score prospects and customers by likelihood to respond to specific offers, and even identify those most likely to become loyal, high-LTV (long-term value) customers.

Data Isn’t Smart
Just because data is big doesn’t mean it’s smart. Until customer data is analyzed, it lacks the ability to predict the future. This is where predictive analytics comes into play.

Predictive analytics uses statistical modeling to identify the unique variables that predict customer and prospect behavior. Isolating the variables that separate customers from noncustomers or responders from nonresponders will improve targeting of new prospects. Predictive analytics can also be applied to customer data, identifying current customers who are at the greatest risk of attrition on the one hand, or who offer the greatest potential long-term value on the other.

There are two main types of predictive models: profile and response. Profile models are primarily used to find prospects resembling your best customers, while response models identify customers and prospects most likely to respond to your offers.

Using predictive modeling, marketers can leverage the vast history of individual transactional data from their own files, along with transactional, demographic and psychographic data from external sources, to predict the likelihood of responsiveness to various promotions. With either an in-house analytics department or an outside partner, retailers can segment their customer audiences and apply the optimal budget, offer and creativity. The resulting audiences can then be addressed through multiple channels, including direct mail, email and targeted display advertising.

To tame big data — and make it actionable — it helps to have data partners that can provide greater perspective on your prospects and customers. The combination of big internal data and relevant external data, coupled with predictive modeling, will enable you to achieve more effective results through segmenting and targeting your audiences.

And that leads to a better ROI: return on insight.

Dave Brown is senior vice president of product management at Wiland, a data and marketing intelligence company. Dave can be reached at dbrown@wiland.com.