Economics & Markets
Does Your Pricing Lock Out the Best Customers?
Netflix's 2011 price restructure cost it about 800,000 subscribers, and the ones who left were its heaviest users.
2026-09-041 min read
A customer's sense of what your product should cost is set once, early, and then defended. It is not recalculated when you add features, absorb higher costs, or discover that the plan was underpriced all along. Economists call this a reference price, and it behaves less like an estimate than like a possession: moving it upward registers as something being taken away rather than as a new offer being made.
Netflix learned the size of that gap in 2011. One subscription had covered both DVDs by mail and streaming for $9.99; the company separated them into two plans at $7.99 each, so a household that wanted both watched its bill rise to $15.98. The economics were defensible and the streaming library kept growing. It did not matter. About 800,000 US subscribers left that quarter, the stock lost most of its value over the following months, and Reed Hastings issued a public apology — then compounded the damage by announcing Qwikster, a separate DVD company, which was abandoned roughly three weeks later.
The detail worth carrying is who left. The subscribers hit hardest were the ones using both halves of the service, which is to say the most engaged customers on the books. A restructure moves bills in proportion to consumption, so the bill that moves most belongs to the person who valued you most. That inverts the usual intuition about churn risk: the customers you would least want to test are precisely the ones a pricing change tests first. A free or cheap entry tier is not dangerous because it forgoes revenue. It is dangerous because it installs a number in your best customers' heads that you will later have to argue with.
Key insights
Customers defend a reference price rather than re-estimating value, so an increase reads as a loss even when the product has plainly improved.
A restructure moves bills in proportion to usage, which concentrates the perceived loss in your most engaged customers instead of your least.
Kahneman, Knetsch and Thaler found that buyers accept increases which cover a seller's costs and reject ones that simply capture more — the attribution decides the reaction, not the amount.
Why it matters
Pricing changes get stress-tested against margin and competitor rates, almost never against the reference price your earliest customers are still holding.
Because a restructure moves the largest bills the furthest, the churn it produces is concentrated in your highest-value accounts rather than spread evenly across the base.
Use this tomorrow
1Pull your ten highest-usage accounts and write down what each one paid in its first month; count how many are still sitting on that original number.
2Take your most recent price change and compute the percentage increase separately for your heaviest user and your lightest user; if the heavy user's percentage is the larger of the two, you have built the Netflix structure.
Go deeper
The mechanism is reference dependence, from Kahneman and Tversky's 1979 prospect theory: people evaluate outcomes as gains and losses measured against a reference point rather than as absolute states, and losses weigh more heavily than equivalent gains. A price rise is therefore not processed as a smaller gain but as an actual loss, which is why it provokes a reaction out of proportion to the money involved. The reference point is simply whatever the customer got used to, which means every introductory price is quietly setting the baseline against which all your later pricing will be judged. Improving the product does not reset it.
A second body of work explains why some increases survive and others do not. Kahneman, Knetsch and Thaler's 1986 paper on fairness as a constraint on profit seeking described a dual-entitlement rule: the customer feels entitled to the reference price, the firm feels entitled to its reference profit, and an increase reads as acceptable when it protects that profit against rising costs but as exploitation when it simply captures more from the same transaction. The distinction lives in the attribution rather than the size, so an identical increase is tolerated under one explanation and resented under another. A restructure invites the second reading by default, because splitting one bill into two makes the increase vivid while leaving the cost story implicit.
