McKinsey now books roughly a quarter of its global fees on outcomes rather than hours. Bain is pushing AI-enabled work toward half its revenue. BCG expects AI work to hit 40% of revenue this year. Read that plainly: the industry that invented the billable hour is quietly walking away from it. Not because clients demanded it first, though 73% of them now say they prefer it. Because the hour stopped being an honest unit of value the moment three associates plus an agent started outproducing a team of ten.
That math is the real story, and it's more uncomfortable than it looks. If an agent compresses two weeks of synthesis into an afternoon, the firm can't keep billing by the hour without either inflating the hours or shrinking the team. Most are shrinking the team. That's the layoff arithmetic showing up across the industry right now — not a downturn, a unit-of-value problem. And once you stop selling time, you have to sell something else: a result, a number, a guarantee. Which sounds like progress. Consultants have complained about the perverse incentives of billable hours for decades — reward efficiency, not inefficiency, finally. Except outcome-based pricing doesn't just change how you get paid. It changes who's exposed when the outcome is wrong.
Here's the tension nobody's pricing model has caught up to. When a consultant bills hours, the client bears most of the delivery risk — they're paying for effort, and effort is verifiable even when the result disappoints. When a consultant bills outcomes, the firm inherits that risk. Fine, in theory — that's supposed to be the trade, more upside for more skin in the game. But the outcome is now produced by a human-agent pair, and the failure modes of that pair are not evenly distributed. The agent can be wrong quickly, confidently, and at a scale no team of ten could have been wrong at in the same afternoon. A bad recommendation that used to take a junior analyst three days to fully develop — and that a manager would catch somewhere in that three days — can now clear every internal checkpoint before lunch, because the checkpoints were designed for the pace of human work, not agent work.
So the firm is asking to be paid like a principal while operating, in practice, like a subcontractor to its own tooling. That's the contradiction sitting underneath the pricing shift: consultants want credit for the agent's speed but haven't yet built the accountability structures that speed requires. You cannot claim a 10x amplification in output as your value proposition and then treat errors as a rounding cost from "the tool." Either the amplification is yours, in which case so are its failures, or it isn't yours, in which case you shouldn't be charging outcome fees for it.
I've watched this play out in real engagements, not hypothetically. A firm compresses a market-sizing exercise from two weeks to two days using an agent, presents it with the same confidence a two-week analysis used to carry, and the client — who has no visibility into how the number was produced — has no way to distinguish a rigorously agent-assisted estimate from a fast, under-checked one. The deliverable looks identical either way. That's the part outcome-based contracts don't yet solve: they price the output, but they don't yet require you to show your exposure to being wrong.
The resolution isn't to slow down or distrust the tools. It's to make the review structure match the new production speed instead of inheriting the old one. If an agent can produce in an afternoon what used to take a team two weeks, the sign-off discipline that used to be spread across two weeks — the second read, the sanity check against a prior engagement, the "does this number survive a hostile question" test — has to be compressed into that same afternoon, not skipped because there's no longer time built in for it. Outcome-based pricing without a matching acceleration of scrutiny isn't a pricing innovation. It's just moving the risk downstream to whoever discovers the error first, and hoping it's you and not the client.
The firms getting this right, from what I've seen, aren't the ones running agents fastest. They're the ones who've decided that amplification and accountability have to scale together — where every agent-produced deliverable carries a visible trail of who checked what, at what confidence, before it left the building. That trail is more valuable to a client than the outcome number itself, because it's the only thing that tells them whether the fast answer was also a checked one.
The billable hour was never a perfect proxy for value. But it did one thing well: it made effort, and therefore accountability, visible. Outcome-based fees strip that away and replace it with a number that looks cleaner and hides more. The consultants who win the next decade won't be the ones who bill for outcomes. They'll be the ones who can still show their work when the outcome is questioned — because the client isn't really buying the number. They're buying the confidence that someone stood behind it before it reached them.
