Subscriptions are a growing business-to-consumer pricing model that has been adopted by industries as varied as microbrewers and razors. Subscription business models provide several opportunities for yield management that non-subscription business models do not. Most significantly, they enable targeted pricing at a customer level on a large scale using observed behaviors and preferences, a practice that can significantly increase revenue and decrease churn typically associated with pricing changes.
Businesses that sell goods or services individually optimize their performance by pricing their products where operating margins are maximized in the current period. For subscription businesses, pricing optimization involves finding the series of prices that maximize the expected lifetime value of a customer relationship over time.
Besides the differences in time horizon, there are different operational limits on pricing strategies between subscription and non-subscription businesses. Non-subscription businesses can segment their market and differentiate pricing across customer groups, but there are practical limits to targeted pricing in their marketing operations. A subscription billing process places few constraints on the pricing strategies of subscription business models, and the additional complexity of targeted pricing is minimal relative to the incremental operating margins created by the pricing strategy.
Subscriber acquisition offers must be optimized much like traditional individually sold products, by finding the price and value proposition that is compelling to the market and balances the quantity and quality of new customers. Limited information on potential customers at the point of acquisition requires pricing strategies to remain less targeted and applicable to a broader section of the market.
Price changes to existing subscribers, often called renewal offers in the publishing industry and rate increases in other verticals, can leverage the additional information collected over the lifetime of that customer’s relationship with the company. This information can include account-specific information such as engagement with the product, complaint calls, payment patterns, geographic region and length of service. Additional information such as approximate age group, education and income level, can be appended to a customer’s data file based on his billing address. Many third party data companies can add information to a customer’s email address or IP address.
Discovering which customer characteristics are important for subscription pricing optimization, and their relative effects on price elasticity, can be approached using two primary methods: analysis of historical customer data and testing of price changes on a representative sample of current customers. Historical data on subscribers can be collected from financial and fulfillment systems, and they often are in the form of transactional data or records of events that have occurred. For instance, financial transactions will indicate when a customer was acquired, made his first payment, changed his billing address and discontinued his subscription.
Fulfillment systems may show delivery times, customer service calls, up-sells or cross-sells offers. These data can be analyzed using a number of statistical regression approaches or other analytical tools. Statistical models estimate how certain factors affect a subscriber’s likelihood of remaining an active customer while controlling for the effects of other factors that may also affect that likelihood. A successful predictive model can support many types of business strategies, including how best to target price changes across the subscriber base.
Validating the predictions from a statistical model can be done using A/B testing, the use of target and control groups to measure the effect a price change or other action has on retention, revenue or other metrics. A/B testing is also able to measure the effect of actions that are outside the observed range of activities captured in the historical data set.
For example, if a company is considering raising their prices above the levels used in the past, an A/B test on a small sample of customers is an excellent way to measure the effect of that price change. A statistical model will not have any data on prices outside of the historical range, so the predictive accuracy may be poor for the proposed pricing action. Ongoing performance reporting can use no-price increase control groups to ensure that the pricing strategy is delivering the revenue expected at an acceptable churn level.
See the table here for an example of a report that indicates the effect of a price increase on a target group relative to a control group.
In this table, the target group received a price increase of $0.61, roughly 5 percent of the pre-increase price level, while the holdout group did not have a price change. The effect of this price change is observed in the number of stops, which increased by about 1 percent, or roughly 5,200 accounts, over the number of stops that would have happened without a price change.
Using these A/B data, we can observe how price changes affect different types of customers. The chart here shows the stop rates across tenure groups following a flat price increase. Customers that have been active on their subscription less than one year have an incremental stop rate of more than 9 percent. Customers that have been active more than five years have an incremental stop rate of less than 1/2 percent.
Combining these incremental stop rates with price data, we can calculate approximate price elasticity by tenure group. The next chart shows the average rates, the average price increase and the calculated price elasticity by tenure level. As is clear in this chart, price elasticity drops for subscribers the longer they are active on their subscription, holding other factors constant.
So, what does this mean for subscription businesses that observe this type of price-response in their subscribers? To maximize their operating margins, it is better to give higher price increases to longer-tenured subscribers and lower price increases to newer customers.
Similar patterns in price elasticity are often observed across income levels, age groups, geographical areas, acquisition channels and payment methods. Data on how these customer attributes affect price elasticity enables a business to find the sequence of pricing offers that maximize the expected operating margins from a subscriber over their lifetime as a customer.
Does differentiated pricing harm the customer? On the contrary, targeted pricing is a net gain for both the business and the customers relative to a one-price-fits-all pricing strategy. Customers that have a low initial value of the product may stop their subscription if they receive a high price increase early in their life cycle before they have the opportunity to become highly engaged with the product. Conversely, long-term customers that place a high value on the product can benefit from product enhancements funded by incremental revenue from new customers retained on targeted prices. In some cases, this type of pricing strategy provides the necessary operating margin to keep the business healthy or to keep them in business altogether.
Subscription businesses are able to apply yield management strategies just as other industries have for many years. Airlines, hotels, and many other businesses apply targeted pricing to maximize their operating margins and customer retention. Subscription businesses need not be any different.
A 20-year veteran in the industry, Matt Lindsay is president of Mather Economics, a global consulting firm that applies a combination of analytical tools and implementation expertise to help businesses develop pricing strategies that maximize operating margins, grow revenue and improve customer loyalty.
