n o ren
Economics & Markets

The Price Change That Reaches Backwards

Unity's 2023 fee reached games already shipped, turning developers' sunk work from a switching cost into a grievance.

A platform whose value comes from what its users have already built into it holds a strange kind of pricing power. Every hour a developer spends learning the tools, every plugin bought, every shipped title that still needs patching, raises the cost of leaving. That accumulated investment reads on a spreadsheet as pricing headroom, because the alternative now carries a migration bill. The mistake is treating headroom as though it were the same in every direction.

Unity tested that in September 2023, announcing a Runtime Fee that charged studios for installs of their games. The number itself was survivable for most of them. What was not survivable was the direction it pointed: the fee applied to titles that had already been designed, budgeted, and released under different terms, so the studios that had committed hardest to Unity carried the largest new liability. Developers who had spent years accumulating switching costs found those costs repriced without their consent. The backlash was immediate and public, studios announced they would move to other engines, and Unity apologized and rewrote the policy inside of two weeks. The fee was cancelled outright the following year.

The lesson is not that price increases cost you customers. It is that a price change reaching backwards converts sunk investment from something that holds people in place into something they resent having made. Subscriptions that raise rates prospectively survive this routinely, because each member re-decides at renewal with full information and nothing already settled is reopened. The same magnitude applied retroactively lands as a broken deal, and it lands hardest on the users most committed to you.

Switching costs are pricing power in one direction only; a retroactive change turns them into an exit motive.
A rise applied at renewal and the same rise applied to past commitments are different products, whatever the arithmetic says.

Pricing power built from switching costs is directional: it exists going forward and evaporates the moment a change reaches backwards.

A retroactive change hits your most committed customers hardest, which is the opposite of how churn normally distributes.

1
List every pricing or terms change your product made in the past year and count how many applied to commitments customers had already made, rather than only to new ones.
2
Take your ten largest accounts and write down, for each, the number of integrations, custom workflows, and shipped products built on top of you; that total, not your feature list, is what your next price rise spends.

Economists describe this through the hold-up problem: when one party makes an investment specific to a relationship, the other party gains leverage it did not hold when the deal was struck. The standard prediction is that anticipating hold-up makes people under-invest from the start. Platforms depend on precisely the investment that hold-up discourages, which is why so much of platform governance is machinery for promising not to use the leverage, in the form of published deprecation windows, versioned APIs, and long-term-support commitments. A retroactive fee is a public demonstration that the machinery is optional.

The reversal is the part worth studying, because withdrawing a price change does not restore the position that preceded it. Unity dropped the fee, but the information the episode released cannot be dropped: developers now know what the company will consider when it needs revenue, and that knowledge prices into every roadmap decision they make afterward. Competing engines gained less from the fee itself than from a year of being the safer default. Trust in this setting is not a feeling but a forecast, and forecasts update on evidence rather than on apologies.