The billable hour is dead. It just doesn't know it yet.
Sounds like provocation. It's math.
For decades every agency ran on the same formula: buy talent, resell it at a markup, keep utilization high. Value = time × rate × utilization. More revenue meant more heads or more hours. That wasn't a business model. It was a law of nature for the industry.
That law had one silent assumption: output is proportional to time. One hour of work, one piece of result. Twice the result, twice the time. AI breaks exactly that proportionality. And with it, the foundation.
Let's do the math
Take a content piece that cost you a day last year. Copywriter, day rate CHF 1,200, one finished piece. You billed the client the day. Your margin sat in the markup.
Today the same writer makes the same piece in 90 minutes. Research, draft, three variants, proofing, AI in the back. Same quality, a fifth of the time.
Now scale it up. Your team ships 200 such pieces a year. Before: 200 days at CHF 1,200, that's CHF 240,000 in revenue from this one service. Capacity was the ceiling - more pieces only came with more heads.
Today the same team makes the same 200 pieces in a fifth of the time. Capacity just grew fivefold. Sounds like good business. It isn't - not in the hourly model. Because now comes the question that decides everything: what do you bill?
You have three roads, and none is comfortable.
Road one: you keep billing the full day per piece. Short term the revenue holds. But you're sitting on a lie with an expiry date. It blows up the moment the client opens ChatGPT and sees their "day project" done in 90 minutes. Then the invoice is gone. So is the trust.
Road two: you bill the 90 minutes honestly. CHF 1,200 per piece becomes CHF 240. Across 200 pieces, revenue from this service drops from CHF 240,000 to CHF 48,000. 80 percent gone. Your freed-up capacity won't save you - you'd have to win five times the work just to break even.
Road three: you push more volume at the old price. But the market won't allow it. When everyone produces five times faster, the price per piece falls. You compete against rivals with the same tools - and against the client who can do it themselves.
That's the trap. In the hourly model, every efficiency gain makes you poorer. The better you get with AI, the fewer hours you can sell. You optimize yourself out of your own revenue.
"This doesn't hit us" - four objections
I know the counterarguments. In order.
"We work on retainers, not hours." A retainer is just an averaged bag of hours. It's priced on estimated effort - and that's shrinking. The moment the client notices the effort behind the retainer is dropping, they negotiate it down. The retainer delays the problem. It doesn't solve it.
"Our clients don't notice." Today, maybe. But your clients are the same people who use ChatGPT privately. The cluelessness has an expiry date, and it's close. Building a business model on the client's ignorance isn't a strategy. It's a bet against time.
"We're premium, we sell quality." Good argument - if your quality were truly untouchable. It isn't, once your client has the same tools you do. AI lifts the baseline for everyone. What was premium yesterday is standard today. Premium remains only where judgment is needed.
"Efficiency is good, we save costs." In the hourly model you don't save costs. You destroy revenue. Your costs are fixed - salaries, rent, tools. When billable hours fall, revenue falls, but the costs stay. Efficiency without a new revenue model isn't saving. It's a leak.
The reflex that doesn't save you
The usual answer: "Then we'll train on AI and get more efficient." That makes it worse. More efficiency in the hourly model means fewer billable hours for the same work. You run faster into the same wall.
What survives
So does everything die? No. Selling time dies. Selling judgment survives.
What used to be scarce is now commodity: production, variants, first drafts. What stays scarce now decides your margin. Asking the right question. Sorting out the mediocre. Keeping a brand consistent across a thousand outputs. Orchestrating impact instead of renting hands. Clients pay more for that than ever, because it's getting rarer.
An example. A small content agency, billed the classic way by effort, margin under pressure. Instead of selling pieces, it flipped its offer. It now sells content systems: brand voice defined machine-readable, quality gates set up, production handed to the client. The client produces themselves - faster and cheaper. They pay for the system and the governance, not the hours. Same people, same craft, different product. Higher margin, no capacity ceiling. That's the jump from time to judgment, made concrete.
What you can do this week
No overnight rebuild. But three steps you can take right now.
First: run your own number. Take your three highest-revenue services. Estimate honestly how much time they really take with AI today - not how much you bill. The gap between the two is your risk.
Second: split your services into two piles. Production that AI makes cheap. And judgment that stays scarce. The second pile is your future. The first, your legacy model.
Third: test it with one client. Sell a result once instead of an effort. A fixed price for a defined impact, not for X days. Watch what happens to your margin.
That's not a threat. It's a fork. There will be two kinds of agencies: the ones that keep counting hours, and the ones that move up the value chain. What that climb looks like is its own chapter - more on that in a coming edition. Here, only one thing matters: the model with an expiry date is the first kind.
Why we can say this so bluntly
Because we stood on both sides. We built, scaled and sold an agency - before this exact shift caught up with the old model. We know the stopwatch logic from the inside. And we know the AI mechanics that dismantle it. This isn't a theory talk. It's a map from someone who has already walked the road.
Take the test
The most honest question this week: how much of what you bill is sold time - and how much is sold judgment? The Model Stress Test gives you the number in a few minutes. It's uncomfortable. But better than the bill the market will write for you otherwise.
