How to spend the right resources on the best customers
Let your resources inform the complexity of your CLV model
How elaborate should your CLV model be? That depends on the amount of funds and talent you can devote to it. Once you’ve identified what resources you can commit, you’re in a better position to determine how sophisticated your model can be.
For example, the analytics team at the online education and training company Career Step uses a handful of signals, such as the product’s purchase price, customer demographic information, and purchase data, to estimate the value of a new customer 13 months down the road. Career Step executives call this their “Day One” metric and laud its accuracy in predicting individual customer value at a particular point in time, as well as its ability to help executives make early decisions about which customers to invest in.
“Before we understood this metric, we sought to grow customer acquisition by simply dumping more money into advertising,” says Ben Woolley, Career Step’s director of marketing for continuing education. “Now that we understand it, we are able to make decisions that grow the business in a smarter way.”
Companies without Career Steps’ data-analysis resources may employ simpler methods, like using purchase price or lifetime purchase value as proxies. For instance, I saw one company that sells single-family-home upgrades use the dollar amounts of its quotes as a way of determining potential value and prioritization of its customers.
If you don’t have the resources for more advanced modeling, consider average revenue per customer, average return frequency, and how many goods or services customers are seeking from your business. Group these customers into as many buckets as you have the resources to analyze, with the simplest groupings being below-average customers, average customers, and above-average customers. Using the data you have, look for commonalities in the data within each grouping. For prospective customers, analyze how their data compares with these established groupings to predict their lifetime value.