On an earnings call in February 2023, Bob Iger explained how Disney had ended up where it was. "We were, as a company, in a global arms race for subscribers," he said — and added that the subscriber count had become the primary measure of success, not only inside Disney but among investors. In the rush to chase subscribers, he said, the company had probably been too aggressive with promotion.
Building that number had cost Disney's direct-to-consumer division roughly $11 billion in cumulative operating losses. But the more interesting thing about the call was what he had just found out.
Two months earlier, in December 2022, Disney had raised the price of ad-free Disney+ from $7.99 to $10.99 — a thirty-seven percent increase, which is not a nudge. And almost nobody left. Iger called the loss of subscribers de minimis, and then said the part that matters: that tells us something.
What it told him was that the base had been more durable all along than the company's own behaviour assumed. Disney had spent years and billions acquiring subscribers on promotion who, it turned out, would have paid more anyway. On the next call he put a name on it — the company had pricing elasticity, and the promotional machine may never have been necessary.
Then, in October 2023, Disney raised the price again, from $10.99 to $13.99. This time 1.3 million core Disney+ subscribers left, a decline the company attributed to the increase and to the end of a summer promotion.
Two increases, two results. The first found headroom nobody knew was there. The second found the point where the base started to give. That is what a stress test produces: not a verdict on the subscribers, but a reading of how much pressure the relationship holds.
Neither answer was visible in the subscriber count, and Disney acted on both — pulling back on promotional acquisition, restructuring the tiers, and reorganising the streaming business around profitability rather than the count.
You don't know what a base is made of until something changes
A subscriber count describes a base under current conditions. It cannot tell you how that base behaves when conditions move — which is the thing that matters, because conditions always move. Durability is the property you actually want to know about, and durability is invisible until the base is put under some kind of pressure.
A price increase is an unusually clean way to apply it. Not the only one — a product change, the end of an introductory offer, a competitor's launch all stress a base in their own way. What a price increase adds is precision: it happens on a date you chose, to a group you defined, and the response arrives in money rather than in a survey.
You ask every subscriber, at the same moment, what the relationship is worth to them — and they answer with money instead of a survey response. Whoever absorbs the increase is worth more this year than last year, from the same person, without you acquiring anybody.
Whoever leaves has told you something too, and it is worth being careful about what. Not that they were a bad subscriber. Someone who paid for seven years and stopped at the new price was enormously valuable, and the reporting was never wrong about them. What they revealed is where their willingness to pay ends — a fact about a point on a curve, not a verdict on a person.
That distinction matters, because durability is not a fixed trait of a subscriber. It is the relationship between a subscriber and a particular set of conditions. The same person can be durable at one price and fragile at another, durable with one product and fragile after it changes. Disney found enormous headroom at $10.99 and the boundary at $13.99, from the same base, fourteen months apart. The base didn't change. The pressure did.
An X-ray of your acquisition strategy
The answer is not the valuable part. What the people who stayed have in common is.
Which channel brought them in. Which offer they took. What price they first paid, and what they were reading in the months before they subscribed. Those are facts about your acquisition strategy, and a price increase grades them in a way no acquisition metric can. It tells you which of the decisions you made three years ago produced subscribers with pricing power today — and therefore where the next acquisition dollar should go, and which channel should stop receiving one. Iger's version of that conclusion was blunt: if they'd have paid anyway, stop paying to get them.
A price increase, read properly, is not a revenue event. It is an X-ray of everything you did to build the base.
You have almost certainly run one in the last three years. The question is whether anyone read it that way.
None of this used to matter as much as it does now. For most of the last decade growth set the value of a subscription business, money was cheap enough that speed was worth paying for, and a subscriber count was a serviceable stand-in for speed. That changed when money got expensive and the question quietly became whether the revenue would still be there in five years. The longer version of that argument is here. The count didn't get worse. It stopped standing in for the thing that mattered.
Somebody ran it on purpose
Disney read its result afterwards, and said so.
This year The New York Times raised the price of its digital bundle from $25 to $30 — not across the base, but for what its chief financial officer described as a cohort of tenured subscribers, who began paying the higher price in the first quarter. He also said the company tends to test all of its pricing, and that this increase followed earlier testing of its own.
So: a designed increase, on a defined group, by a company that runs them routinely.
What makes a price increase informative is not its size. Disney's was thirty-seven percent and produced almost no departures, which was itself the finding. What matters is that it forces a decision, and that the cohort is defined before the notice goes out.
On that second count this one is well built. Not new subscribers on introductory rates, whose behaviour tells you mostly about your offer. Tenured subscribers — people who have already stayed, already renewed, and already look durable on every conventional measure. They are the part of a base a company is most confident about, which makes them the part it knows least about, because the confidence has never been tested.
And the Times is the company the industry watches. It is closing in on 15 million subscribers, a target treated as the definition of winning.
As reported for the quarter ended June 30, 2026.
What it revealed
Digital-only ARPU rose 3.1 percent, which the company attributed to subscribers moving off promotional pricing and to the increase on tenured subscribers. Management presented it as a success, and made no mention of elevated churn.
Now try to use that number.
A 3.1 percent move is consistent with the increase going beautifully — tenured subscribers absorbing it almost entirely, the average held down only by the 1.5 million new subscribers who joined during the year at lower prices. It is equally consistent with the increase flushing out a meaningful slice of the tenured base, because losing subscribers who were paying $30 pulls the average down from the other direction.
Those are opposite findings. The number reads identically either way — which means the company's characterisation of the increase as a success is not something the figure could contradict, whichever one actually happened.
Nor can you get around it. The Times reports net additions — joiners minus leavers, never separated. The 280,000 could be 500,000 people joining and 220,000 leaving, or 1.1 million joining and 820,000 leaving. Same headline, two entirely different businesses. Nowhere in the release can you find how many subscribers cancelled, how long they had been subscribers, or which offer first brought them in.
And the reporting is getting less granular, not more. In February the company said it would stop breaking out subscribers and ARPU by subscription type, reporting only total digital-only subscribers and total digital-only ARPU going forward. Whatever the reasons — and there are ordinary ones — the bundle, the single-product subscriptions and the promotional cohorts now arrive in the same two numbers.
Disney, by contrast, told everyone. A number, a direction, and a cause: 1.3 million core subscribers gone, attributed to the price increase and the end of a promotion. You can disagree with the attribution, but there is something there to disagree with.
So the most informative event in the Times' subscription year — and I would defend that description, because no product launch, bundle change or campaign tells you as much about a base as watching it decide whether to keep paying — happened, produced an answer, and none of it survived into the reporting.
This is not a company being cagey. The Times discloses more than most publishers report at all: digital-only ARPU, the split between bundle and single-product subscribers, a chief financial officer who discusses mix openly on the call. If reporting at that level cannot tell you what a price increase revealed, no earnings release will. This is not information anyone publishes.
Fifteen million of what
The target was never wrong. It was incomplete.
Fifteen million subscribers who absorb a price increase and fifteen million who don't produce exactly the same headline, on exactly the same schedule, with exactly the same feeling of progress. They are two different companies. One of them can raise prices next year. The other can only hope nothing changes.
Every publisher chasing a subscriber number is chasing one of those two. The subscriber number cannot tell them which.
This is worth saying plainly, because it reframes something a lot of subscription teams have been carrying privately: when the count keeps climbing and the business doesn't feel any stronger, the instinct is that something must be wrong with the execution — the funnel, the offers, the retention programme, the team. Sometimes it is. But no amount of execution can make a subscriber count tell you something it was never built to measure. The number cannot say whether the base is getting more resilient or less, and it never could.
There is one question that separates them, and you can ask it about your own business this week, because you have almost certainly run the test.
Pull the two groups from your last price increase — the ones who stayed and the ones who left — and break them out by acquisition channel, introductory offer, tenure, and what they were paying before. Find what the people who stayed have in common. Then ask the question that actually costs money: are you still buying more of them?
That is the X-ray. Not what the increase did to revenue — what it revealed about everything you did to build the base, and what you should be doing next.
It is worth doing for its own sake. It is worth doing sooner if there is a financing, a sale or a board review anywhere on the horizon, because the same analysis will be run by somebody eventually. The only variable is whether you run it first, while there is still time to act on what it says.