Businesses have been segmenting customers for years, but the era of Big Data is making it more essential -- and complicated -- than ever. The Big Data challenge is not simply a race to accumulate information; it is a race to understand customers more intimately, and to act on that knowledge.
Segmentation is a foundational element of understanding customers. In its simplest form, customers are grouped based on similar characteristics. As the data improve (demographic, attitudinal and behavioral), the approaches to segmentation become more sophisticated. Yet many businesses suffer from a kind of segmentation paralysis.
If you ask an executive or marketer, they will tell you that knowing the customer is the key to success. Yet if you look more closely, you often hear: "We have lots of segmentations, but to be honest we just end up filing them away." Or, "Our segmentations don't really mean anything. They are all are pretty much the same." And, "We use it as a starting point for setting strategy, but never refer back to it going forward."
The paralysis often results from one of the following approaches:
The Marketing Study. This is segmentation that is owned and commissioned exclusively by the marketing department. It has not been built to answer a question, but rather to describe high-level generalizations of the customer. It doesn't account for the variety of key questions, objectives or insights needed for the entire organization. Great segmentation starts by identifying what the business, not just the marketing department, wants to know about its customers and how customer insight could make business decisions more effective.
Someone Else's Segmentation. This is an off-the-shelf generic segmentation that describes customers based on widely available data. These segmentations are not unique to a specific business and do not describe the behaviors or attitudes with which people think about your business. Businesses need to start with their own strategy and the right data. If the data are lacking, scoring every customer into segments may be impossible, but ensuring that the majority are covered is critical.
Inside-Out Segmentation. Here, segmentation is approached through profit and revenue, then used to characterize the customer, rather than accounting for how customers actually behave. This approach is common but becoming increasingly outdated. Big Data has made it easier for businesses to characterize each household based on its behavior, which ensures precision and relevance. Behavior is how a customer acts, not what he says, where he lives or how you think he acts. Don't build a segmentation based on spend alone, base it on the behaviors that drive the spend.
Average Segmentation. This segmentation is built to account for all types of behavior and data, also known as the "one-segmentation-to-rule-them-all" challenge. This segmentation often leverages the wrong data, such as demographic data, that can simplify customers and their behavior. Ultimately, this produces a model in which the segments and the descriptions of customers are too similar, and it becomes impossible to act on them. Instead of a single segmentation, a portfolio of segmentations that seek to solve different business issues is often required. For example, a pricing problem will often need a pricing segmentation; a loyalty program will need a separate strategy.
The Shipping Tanker Segmentation. This is where stability has been over-engineered into the model and therefore nothing changes. Customers are well segmented, but changes in customer behavior are not accounted for. Finding the right balance is one of the most difficult areas to manage within segmentation. Rather than thinking exclusively about segments, think about the set of attributes that rely on continuous scores against a specific behavior or attitude. This provides greater flexibility to grow an understanding of the customer over time and apply it across all dimensions of the business to drive parallel execution rather than limited to a major event.
The Einstein. This segmentation that is complicated and uses so much data and unknown statistical techniques that no one can explain or leverage it. Often these segments are developed exclusively using complex modeling. When this segmentation is socialized throughout the organization, buy-in is difficult, as the complexity produces a level of distrust. Good segmentation allows you to develop clear customer segment strategies and should be intuitive enough to ensure that the way the segments fit together makes sense for the entire organization to act on and measure against.
Big Data gives us the opportunity to move away from a one-size-fits-all approach and allows us to manage and activate on many segmentations with the consistency required for long-term analysis. Companies that view segmentation as the enabler of change and can move away from the "descriptive data dump" to develop a deeper understanding of customers, will soon outpace their competitors.