When a resource gets cheaper to use, people use more of it, not less. For agencies, it is why cheaper AI output tends to create more work, not free up the calendar.
Also known as Jevons effect, rebound effect
Jevons Paradox is named for the economist William Stanley Jevons, who noticed in 1865 that more efficient steam engines did not reduce Britain’s coal use. They made steam power cheap enough to put in many more places, and total coal consumption climbed. Efficiency lowered the cost of each unit of work, and demand for that work grew faster than the savings.
In AI, the resource is usually tokens, compute, or simply the time it takes to produce a draft, an image, or a variant. As models get faster and cheaper per task, teams do not bank the savings. They run more prompts, generate more versions, and ask for more deliverables. Total spend and total volume often rise even as the unit cost falls.
Most agencies adopt AI to save time and money. Jevons Paradox is the reason those savings so often fail to show up where leadership expects them.
Scope expands to fill the new capacity. When a first draft takes ten minutes instead of two hours, clients and account leads ask for ten concepts instead of three. The hours saved per asset get spent on a larger number of assets.
Pricing and margins get squeezed from both sides. If clients see that production is cheaper, they expect lower fees and more output at once. An agency that sells hours has to rethink what it is actually selling. The Creative Cadence Framework lays out how to move from billing for time to being paid for impact.
Review becomes the bottleneck. Cheap generation moves the cost to judgment: strategy, taste, brand checks, and approvals. More volume means more to review, and that work is still done by people.
A mid-size agency rolls out AI tools for paid social. Resizing and copy variants that used to take a designer and a copywriter a full day now take an hour. Within a quarter, the client’s media team asks for forty variants per flight instead of eight, because testing more creative is now “free.” The agency’s production hours barely drop, its API bill grows, and the account manager spends Friday afternoons sorting through variants nobody has time to review. The fix was not less AI. It was setting a cap on variants per flight, agreeing on a review step, and repricing the retainer around outcomes instead of hours.
The automations and agents module of the workshop teaches you how to build AI workflows that compress the busywork without taking the craft out of the studio.