Life Support For Subscribers - using 'survival analysis' to help publishers build their subscription strategies [insight]
In 2002, Knight-Ridder, then the second-largest newspaper publisher in the U.S., hired me as a consultant to analyze its subscription-pricing strategy. At the time, circulation and revenue were declining, despite the company’s aggressive discounts.
To kick things off, Knight-Ridder sent me data from one of its newspapers, the St. Paul Pioneer Press. As soon as I plotted the data on newspaper renewals, I got a case of economic déjà vú. The numbers were remarkably similar to data I had seen on patient lifespans in a course in the economics of healthcare.
That epiphany eventually became the bedrock of my consultancy, Mather Economics, a 40-person company that analyzes customer data for 500 publications in 12 countries. Every week, clients provide us with information on 30 million households with subscriptions worth $4 billion annually.
What do subscriptions have to do with survival rates? The similarity I noticed allowed me to unlock insights using a field of statistics known as “survival analysis.” It was developed to analyze patient data and calculate how specific treatments would affect lifespans. The same type of analysis is used to predict failures in machinery. In applying this approach to publishing, I wanted to predict how long an individual subscriber would remain active as a customer under particular pricing strategies, retention campaigns, and product-bundling offers. The failure event in healthcare economics is a patient’s death. In my model, it was a canceled subscription.
We found that price increases have a significant effect on a subscriber’s likelihood of cancellation, but that effect varied dramatically across the customer base. Certain customers were 10 times more likely to cancel their subscriptions after a price increase. It quickly became clear that a one-price-fits-all model was a poor strategy for publishers — and that determining an individual customer’s price sensitivity allowed the development of a more effective one. Much like the airlines, publishing could use sophisticated pricing analytics to increase revenue and operating margins.
To see how variable subscription pricing helps publications, just compare media in the U.S. and Europe. European publishers still haven’t adopted individualized pricing, partly due to data privacy laws that make the required analysis difficult. As a result, according to a World Press Trends report, from 2011 to 2015 news media companies in Europe lost 21.3 percent of their print circulation base; over the same period in the U.S., the loss has been 8.7 percent. Several factors contributed to the gap, including European publishers’ greater reliance on single-copy sales, but the use of customer data by American publishers accounts for much of the difference. Had U.S. publishers followed the European strategy during that period, they would have lost an additional 5 million newspaper subscribers, at a cost of more than $1 billion in annual revenue.
Many factors that affect subscription cancellation rates seem obvious. As subscribers’ income goes up, for example, their sensitivity to price goes down; most people who make more than $100,000 a year won’t think twice about paying a few more bucks a year to get a newspaper or magazine. For subscribers who are less well-off, it’s a different story. That’s one reason why, at the end of the four years, there are about twice as many high-income subscribers remaining in a cohort as low-income subscribers.
Income isn’t all-important, though. The method an individual uses to subscribe — a publication insert, direct mail, a telemarketing call — is at least as significant a factor. A big reason so many low-income subscribers cancel is because many of them subscribe via channels that aggressively push low-priced offers. Given their financial straits, it should be no surprise that they cancel when those offers expire. But the same applies to high-income subscribers. They are 50 percent less likely to cancel if they bought a subscription from direct mail rather than from a high-pressure method such as door-to-door sales.
Demographics and acquisition channels are only a couple factors we use in our print subscription models. With digital subscriptions, the variables increase substantially. We analyze subscribers’ website visits, articles read, videos watched, average time per visit, and time between visits. We also look at comments, social media shares, contest registrations, survey responses, and many other variables.
Successful offers usually combine a compelling price point — such as 10 weeks for $20 — with a long initial term, an automatic renewal, and a recurring credit card payment. But there are often counter-intuitive results, which means the task of developing and testing offers is never done.
Publishers can reduce subscription churn by 15 percent within a few weeks of starting a data-driven retention campaign, and the numbers usually get better over time. Pricing strategies can reduce price-related customer losses by as much as 75 percent. And as publishers begin to use customer analytics and big data, they often find that improving retention offers the greatest return on investment because saving an existing customer is far cheaper than acquiring a new one.
When print newspaper advertising was at its peak, many publishers thought these tactics smart but unnecessary. A few years later, when ad revenue plummeted, their perspective changed. “It is incredible how much money we left on the table all of those years,” one executive told me. “We could have been charging so much more for our product.”
Written by Mather Lindsay, the president of Mather Economics, a global consulting firm. Over the past 15 years he has developed pricing strategies and predictive models for clients including Gannett, Tribune, McClatchy, Dow Jones, and the New York Times.