Fable
A New Year’s dish eaten for two years: fragrant the first year, bland the second
In the south of the city there was an old restaurant that every twelfth lunar month offered a "New Year's table" subscription: a household paid a monthly share, and the restaurant delivered a table of New Year dishes to their home each month. In the first year, thirty households signed up, each paying forty taels per month. The accountant took out the abacus, gave it a flick, and said this year's ARR was three hundred eighty-four taels — thirty households times twelve months times forty taels, a beautiful number.
By the second year, the restaurant had developed twenty new customer households and signed the same table packages. Unlike the first year, five old customers canceled because they found the dishes monotonous, three upgraded from one table to two, and the remaining twenty-two stayed the same. The accountant worked the abacus again and said second-year ARR had risen to four hundred eighty taels—fifty households subscribed. Outsiders hearing this number applauded and said it had doubled.
The real accounting was done by the old cook. He took out a detailed ledger and singled out the thirty households that signed up in the first year. The five that canceled went to zero, the three that upgraded paid eighty taels each, and the twenty-two that stayed paid forty taels each. Adding it up, the annualized revenue these thirty households generated by the second year was three hundred eighty-eight taels—only four taels more than the first year's three hundred eighty-four. The old cook said this number was called NRR, and it had just passed one hundred percent.
The shopkeeper only calmed down after looking at the detailed accounts. ARR rose by nearly a hundred taels because new customers increased by twenty households; but old customers' NRR barely held—the three upgraded households propped up the hole left by the five churned households, but there was almost no growth. If old customers continue to churn in the second half of the year, ARR will also be dragged down, because the acquisition cost of new customers is much higher than the maintenance cost of old customers.
The old chef added another layer. He separated thirty old customers by signing channel—of the nineteen referred by neighbors, only one churned; of the eleven who found them through flyers, four churned. The neighbor group's NRR was nearly twenty points higher than the flyer group's. Only then did the shopkeeper realize that ARR will not tell you which channel's customers are the most stable.
At the year-end review, the restaurant's two accountants reported results at the same time. One said: This year ARR rose 25 percent. The other said: Existing-customer NRR fell three points, and new customers made up the difference. The owners listening to the report discussed the first sentence enthusiastically and fell silent at the second.
Later, two numbers were placed side by side on the restaurant's report: the ARR on the left showed how big the overall pie was, and the NRR on the right showed whether the pie was leaking. New restaurants liked to stare at the number on the left; old restaurants felt the one on the right was the real conscience.