"Only pay for results" has been a marketing tagline for two decades. It almost never held up. Agencies that promised it either quietly went back to retainers, cherry-picked deals that were going to close anyway, or went out of business. So it's fair to ask what's actually different now — and the honest answer is not the sales pitch. It's the cost curve.
Why it used to fail
An old-line agency is a labor business. To deliver, it pays people — strategists, writers, media buyers, analysts. Betting on outcomes meant gambling those salaries (and often the ad budget) against an uncertain return. Win and you share the upside; miss and you've burned a payroll you can't get back. No firm can survive many of those misses, so the rational move was to de-risk with a retainer and hand the uncertainty to the client.
That's not greed. It's arithmetic. When delivery is expensive, betting on outcomes is reckless.
What changed
AI collapsed the marginal cost of the work itself. Research, prospect targeting, copy, sequencing, creative iterations, reporting — the parts that used to consume the most hours are now mostly software with a human steering. Delivery cost didn't drop a little. It dropped enough to change the category of decision.
When the cost of a miss approaches the cost of compute, you can finally afford to be wrong sometimes.
That's the whole unlock. When each bet is cheap to make, you can make many of them, absorb the ones that don't land, and let the winners pay for the entire portfolio. It's the same shape an investor's book takes — a few outcomes carry the fund. A retainer agency can't run that book. A software-cost one can.
The thing the math doesn't fix
Cheaper delivery makes the bet survivable. It does not make every bet good. The one variable AI doesn't touch is whether your team can actually close the at-bats we create — and that sits entirely on your side of the table. Which is why, even with the economics working, selection still does most of the heavy lifting. The cost structure lets us bet. It doesn't tell us who to bet on.