In the age of analytics and Big Data, all customers are not created equal.
Successful companies have believed that maxim all along. Some have even acted on it. The idea that all customers should be treated like kings, regardless of their bottom-line value to the seller, is simply not tenable.
Now comes a flood of data and the newfound power to assess customer value through sophisticated analytics. And sellers find themselves with previously unattainable insights into how buyers prefer to interact with them.
For customer segmentation to work, certain pieces of key information must be gleaned at the outset. Earl van As, vice president of marketing and product management with Conexiom, a provider of systems for automating sales orders and invoices, lays out three crucial categories of customer intelligence: purchase patterns and future demand, preferred means of making purchases, and time spent with customer service reps.
The ultimate aim, says van As, is to assess the bottom-line value of each customer: which are the biggest “profit makers,” and which account for the highest cost to serve.
Conexiom draws its insights from the order data that it processes on behalf of wholesale distributors. According to van As, many businesses approach customer segmentation from the standpoint of account location, order quantity and season. While those are all valuable bits of data, companies often ignore the equally vital details that calculate the effort and total cost associated with the processing of an order.
Van As cites the case of a large maintenance and repair company that believed its customer-satisfaction levels to be comfortably high. When it undertook an analysis of how buyers were actually purchasing, however, it discovered an opportunity to convert time-consuming e-mails into automated orders.
“Change doesn’t have to be difficult,” says van As.
“Don’t think of it as a big enterprise-level initiative. You can smart small.”
It can be as simple as setting up rules in a exchange server — for example, flagging any communication that comes in with the word “order” in the subject line. Customer service and inside sales are especially valuable places in which to smoke out the manual entry of orders, van As says.
The degree to which customers are taking up the time of customer service reps is a crucial metric for enabling effective segmentation. According to van As, many organizations look only at the final data set related to an order, instead of examining who was involved in processing that order, and figuring out the real cost associated with it.
Plenty of companies can tell you how many customer-service reps it has in place. (Those that can’t are in even deeper trouble.) They can even track the percentage of CSRs’ time spent on processing data. But too few can say how many orders they receive by email from a given customer.
“Just knowing that a customer is sending you 100 orders a month doesn’t tell you a lot,” says van As. “You may have insight into when they might order, but not into the cost structure associated with entering that order.”
Having the right information in place is only the first step toward true customer segmentation. The seller must be able to translate raw data into reliable predictions of future customer demand and behavior. Companies should be able to combine prior knowledge of what customers will buy with how they will place orders. Only then can they allocate resources appropriately, and improve margins by focusing on the most valuable accounts.
To a great extent, the right decisions will be based on historical data related to past purchases. But the calculations can get tricky where new product introductions are involved. Such items have no history of demand, so companies find themselves looking at prior sales of similar products, and hoping that the results can be applied to the new offerings.
The time horizon for making accurate predictions will vary according to the product in question. A distributor of air conditioners, for example, can reliably expect customers to behave in a certain manner during the summer months.
Extending predictions out over multiple years can be far more difficult. Van As says it’s important to integrate analytical events with a formal sales and operations planning (S&OP) process, which is designed to make sense of future demand over longer periods of time.
Of course, no prediction will ever be 100-percent accurate, and the longer the time horizon, the greater the likelihood of error. What’s more, purchasing patterns can change, meaning that companies must be prepared to pivot with every shift in market demand and level of automation. Each prediction is by definition provisional.
Nevertheless, the huge amounts of data available to sellers today, coupled with analytic prowess, can provide an unprecedented degree of visibility to transactions and customer behavior. In the process, it can engender true customer segmentation. “Customer-driven” is the mantra of just about every company today. What businesses need to determine, however, is exactly who is doing the driving.