An agent just did in twenty minutes what you would have scoped as three days. Pick the number that stings more: the three days you can no longer bill, or the margin and firm valuation that quietly leave with them.
That is why outcome-based pricing is suddenly the only conversation in professional services. The billable hour is dying, AI is the cause, and everyone can feel the revenue draining out of the old model. TSIA calls it the cannibalisation dilemma: the more efficient you get, the more of your own hour-based revenue you erase. Forrester says consultancies are already putting fees at risk, with PwC reportedly doing it on around a third of engagements. The prescribed fix is near-unanimous: stop selling hours, start selling outcomes.
It is a good argument. It is also aimed at the wrong layer.
The billable hour was never just how you charge. It is the unit of account for the whole business. It is how you estimate, how you cost, how you schedule, how you forecast revenue, how you measure utilisation, and, when you sell the firm, how a buyer values it. Almost the entire conversation about professional services pricing models is happening at the invoice, the visible surface, because that is the easy thing to see. Outcome pricing changes that one number. It leaves the other five still denominated in hours.
The billable hour pricing model is load-bearing in four places, not one
Here is the part the pricing discourse skips. You can change how you charge in a quarter. A new master agreement, an outcome clause, a shared-risk schedule, all of that is a contracting exercise. It is real work, but it is bounded.
Re-denominating the rest of the business away from hours is not bounded. It is years of work, and most firms have not started.
Walk the firm and the hour shows up in four distinct jobs:
- Estimation. You scope a deal by guessing how many hours each role spends, then multiply by a rate. Effort times rate equals price. Every estimate you have ever produced runs on that arithmetic.
- Capacity and scheduling. You plan delivery in person-hours and person-days. Utilisation, bench, who is free in March, all of it is hours against available hours.
- Revenue forecasting. You forecast recognition off scheduled effort. Hours booked, hours to deliver, hours times rate landing in a given month.
- Firm valuation. When you sell, the buyer reads your revenue per head, your utilisation, your effective rate. The hour is baked into the multiple.
Outcome pricing fixes the first column of the first job and nothing else. You can write a fixed-outcome contract on Monday and still be estimating, scheduling, forecasting, and valuing the firm in hours on Tuesday. The invoice changed shape. The substrate underneath it did not move.
Where everyone is looking versus where the problem sits
The reason the whole industry is talking about outcome pricing is that it already knows the hour is losing meaning. That instinct is right. An agent that does in twenty minutes what billed at three days makes the three-day estimate a lie, and everyone can feel it. So the natural move is to stop selling the three days and sell the result instead.
But notice what that does and does not solve. Selling the result protects your top line from the deflation. It does nothing for the four jobs underneath, because all four still ask the same question: how much work is this, and what does it cost us to deliver? Outcome pricing changes the answer the buyer sees. It does not change the question your own business has to answer to stay profitable.
This is the quiet trap. A firm reprices the top layer, declares itself outcome-led, and keeps running its estimating sheet, its capacity tracker, and its forecast on hours that no longer mean what they used to. The price is now a promise about a result. The cost model underneath is still a guess about effort, and the effort number is exactly the thing AI has made unstable. You have transferred delivery risk onto yourself (that is what outcome pricing does) while keeping a cost engine that cannot see the risk you just took on. That is not a pricing problem. That is a margin leak with a delay on it.
Estimation: the hour was always a proxy, now it is a bad one
Estimating in hours was never really about hours. Hours were a proxy for cost. A senior developer at a day rate, multiplied by the days you think the work takes, gives you a cost base you can put margin on. It worked because, for decades, headcount was a clean proxy for cost. More work meant more bodies meant more hours meant more cost.
Agent-plus-human pod delivery breaks that proxy. The pod that ships the feature might be two people and a fleet of agents, doing the scope that used to need six people. The cost base is no longer a clean multiple of full-time staff. Mphasis, one of the cleaner public examples, reports doing the same scope with fewer people, citing moves like a hundred-person team down to sixty. If you estimate that engagement in person-hours, your number is wrong in a direction you cannot see, because the relationship between the work and the bodies has changed.
So estimation has to model two things it mostly ignored before. It has to carry a blended cost base where the mix of human time and agent-assisted output is explicit, not assumed. And it has to hold price separately from effort, because the price is now an outcome and the effort is now a variable you are managing down. The hour as the single unit cannot do that. It collapses both into one number and hides the gap.
Capacity and forecasting: the two jobs repricing forgets entirely
Capacity is where this gets expensive fast. Capacity planning is hours, full stop. Demand in hours against available hours, by role, by practice, across signed and pipeline work. When the deal is priced on outcome rather than hours, the capacity question does not go away. It gets harder. You still have to staff the work. You still have to know whether you can deliver the next three signed outcomes without burning the team or hiring ahead of revenue. But the contract no longer tells you how many hours the work is, because you sold a result, not a quantity of time.
So you need a capacity model that derives effort and resourcing from scope, independently of how the deal is priced. The price says "outcome". The capacity model still has to answer "how many of which people, when". If those two are not linked through a shared model, the outcome price and the delivery plan drift apart, and you find out at the worst possible time, mid-delivery, with the margin already committed.
Revenue forecasting has the same shape. Recognition is read off scheduled effort and milestones. Switch to outcome pricing and the cash now lands against results and acceptance, not against hours logged. If your forecast still smooths revenue across booked hours, it is forecasting a thing the contract no longer pays for. The recognition timing has to come from the deal structure, the milestones and outcomes, not from a parallel finance model still counting hours.
Neither of these is solved by changing the invoice. Both are solved by having a commercial model that carries scope, cost, capacity, and revenue timing together, so that when the price moves to outcomes, the layers underneath can still answer their own questions in their own terms.
Be honest about the timeline
None of this is as far along as the louder voices suggest. HFS, which coined "Services-as-Software" and has every reason to push the thesis, says plainly that "non-linearity is not yet broad-based". Across twenty-five-plus major providers in 2025, revenue per head rose around 1% year on year and margin per head around 3.5%, and the gains sit in maybe five to ten firms, partly driven by restructuring rather than AI. The contrarian read, from Simon-Kucher, is that the billable hour is "mutating, not dying" and "nobody has cracked the nut yet". BCG argues the whole shift is net expansionary, up to $200 billion of new integration work, because someone still has to wire the agents into messy enterprise systems.
So this is a transition, not an extinction, and the timeline is genuinely contested. That cuts in an uncomfortable direction. If the shift were instant, you could rip out the hour and rebuild around outcomes in one move. Because it is gradual, you spend years running both at once: some deals on hours, some on outcomes, the same delivery teams, the same capacity pool, the same P&L. The firm that can hold both unit models in one coherent view through that transition is the one that keeps its margin legible. The firm that reprices the invoice and leaves the substrate alone spends those years unable to explain why the numbers stopped reconciling.
What this means for the model underneath
The firms that survive this will not be the ones that reprice fastest. Repricing is the easy part, and the part everyone is already doing. Choosing the pricing model is a real decision, but it is bounded. The hard part is keeping scope, cost, capacity, and margin visible when the hour stops being the unit that ties them together.
That is a tooling problem before it is a strategy problem. It means a commercial model where price and effort are separate dimensions, not one number, so an outcome price can sit on top of a cost base built from a blended human-and-agent mix. It means capacity that derives from scope rather than from the contract's pricing, so you can still staff an outcome deal. It means revenue recognition read from deal structure rather than from hours logged.
This is the part of Estii that the shift makes more relevant, not less. A deal in Estii carries seven dimensions in one model: scope, price, margin, schedule, capacity, revenue timing, and handover. Price is governed separately from effort through rate cards, so a deal can be priced to an outcome while the cost base underneath stays explicit. Adjust scope and effort, value, price, and the delivery schedule recalculate together rather than drifting apart. Forecasts reads capacity at confirmed, forecast, and exposure levels and recognises revenue against milestones, so the schedule and the cash come from the deal structure, not from a separate hours-based sheet. The point is not that Estii prices outcomes for you. It is that the model does not assume the hour is the only unit, which is exactly the assumption the rest of the stack is about to lose.
The invoice is the easy thing to change, which is why the whole industry is changing it first. The unit of account is the hard thing, which is why almost no one has started. The hour stopped being a good proxy for cost the moment one engineer plus a fleet of agents could deliver what used to take a team. Repricing hides that for a quarter or two. It does not fix it. The firms that come through this will be the ones that can still see scope, cost, capacity, and margin once the hour underneath them stops meaning anything.
