Why Data Mining Is Key to Successful Segmentation

Are you getting the most out of your campaigns? If not, your next step could be to go beyond segmentation, to look at deeper insights from data analytics and data mining.

Successful marketing campaigns are predicated on several components: targeting the right audience, at the right time, in the right channel, with the right message. However, for a marketing campaign to be successful, a marketer first has to identify an opportunity, either a growth opportunity or a negative trend to be reversed or decreased. The marketer then needs to create a campaign and define what success is, then decide how to measure it.

These steps require powerful analytics, including data mining, segmentation, contact strategy and planning, incrementality analysis, and analyzing campaigns in hindsight. The starting point for all of these is segmentation.

Data Mining and Segmentation: Reinventing Your Strategies

Could you be overlooking key considerations in your segmentation? Many marketers, caught up in immediate needs and questions, miss important insights. How does this happen?

First, remember that there are two types of segmentation. There’s the everyday, operational one aimed at very clear, understandable tasks such as launching a new product or creating a special promo. The advanced, analytical segmentation is a part of data mining, and machine learning (ML) is rapidly increasing its effectiveness, as well as promoting intense interest in its use.

With personalization and machine learning models, traditional contact strategies built on basic segmentation methodologies like recency-frequency-monetary (RFM) analysis are losing popularity, replaced by triggered campaigns and other automated tools. However, even in today’s world of extensive digital experience, there is a place, a time, and a good use for traditional or large-scale segmentation.

How do marketers build relevant segmentation strategies? There is no one right answer; there are actually many. But there is a common link between all successful segmentation strategies. It all starts with the insights gained from data mining.

Cluster analysis is a common tool, especially in the presence of extensive transactional and/or sociodemographic data. Even basic groupings of future lifetime value increments have powerful applicability.

Data mining may occasionally seem like feeling your way around in the dark. However, these approaches provide unexpected insights that can become a competitive advantage for an organization. They can identify challenges for an organization that may not be immediately apparent. Examples include increase in lapsing rate, a disproportionate high percentage of low-value customers, low customer lifetime value (CLTV) in certain acquisition channels etc.

These negative trends may be masked in high-level reporting. For example, an increase in low-value customers may be masked by an increase in spend by higher value customers. Low-value customers generally have a higher probability of lapsing that will eventually appear in customer universe reporting. However, by then, you might have already lost an opportunity.

All of these types of analyses often lead to a better understanding of the customer universe and the business. Coupled with customer feedback, these can become the foundation for developing a segmentation strategy. In the above example, average order value (AOV) will be a critical component of segmentation strategy, especially changes over time. This will segment customers based on the challenges the company is facing.

Other ways to design a company’s segmentation are purchasing and style preferences (especially in apparel, jewelry, home goods etc), pricing/promotion preferences etc.

However, segmentation analysis by itself is incomplete. It is nothing more than raw analytics that has not been translated into action.

What is missing is a strategic goal and a contact strategy associated with each segment. Every segmentation strategy should include a segment of “New.” New customers strategy should be onboarding on the brand with a clear contact strategy aligned with exposing new customers to the capabilities, positioning and ethos of the organization.

Similarly, other segments should have predefined goals and the associated strategy. The top customer segment (tier 1, best customers, however they are named) strategy should be “keep them there” with an associated contact strategy. Let’s for a minute leave UGC, influencer status and brand ambassador statuses aside.

Similarly, for other segments that may involve increase in transactions, increase in AOV, expand dept penetration, update to higher margin categories, and more.

This type of segmentation works by grouping customers in large buckets (ranging from 2-10 segments, rarely more) and may appear antithetical to the world of customized messaging and triggered campaigns. However, these two approaches are in fact, complementary, so long as the appropriate prioritization exists. All customers should get a cart abandonment contact. That does not conflict with the opportunity to notify the top customers that they have a sneak peak at the new collection, open only to best customers or a special promotion in key departments again open to best customers only.

Are Your Customers Exhibiting an “Off Shopping Pattern”?

Sometimes, tracking customers’ movement within segments creates a new opportunity to present custom messages to get the movers back to the higher value segment. It’s also an opportunity to collect data that could point to the reasons for the move. Of course, it’s better to find out from this group, rather than wait until they lapse or unsubscribe to try to find out

What Have We Learned about Customer Analytics?

Data mining is an integral part of determining the strategic segmentation for the organization to supplement triggers based on specific and recent actions. This strategic segmentation allows direct-to-consumer (DTC) companies to identify the appropriate strategies to fit specific customer groups, to accomplish longer term goals (like improved retention) rather than shorter term goals (convert an abandoned cart).

CDPs are key for helping you accomplish this. CDPs with imbedded analytics and data mining built on their inherent capabilities in unified view of the customer to allow marketeers to easily and efficiently (read: code-less!) navigate their wealth of customer data to identify the segmentation strategy that best fits their needs.

And this segmentation needs to be periodically reviewed, adjusted and expanded based on new data mining insights that should be an ongoing process as well as new data sources that may further refine the segmentation strategy. Strong CDPs allow for that flexibility to incorporate new data as well as monitor a segmentation strategy longitudinally and refine over time.

For some good ideas to use on segmentation, check out this behavioral segmentation article or check out this valuable Retail Touchpoints report on successful segmentation and loyalty strategies.