A consultancy I know won its first big fixed-outcome deal last year. Not time and materials, not a day rate with a cap. A single number for a defined result, signed. The room celebrated, because the number was good and the win was the kind that gets mentioned in the next board update.
Then delivery started. The work ran about fifty percent longer than the estimate that won it. Nobody had lied. The estimate was simply the number that cleared the deal, produced the way every estimate before it had been produced, by people whose job was to get the work in the door. Under the old model that optimism was free. This time the overrun came straight out of the margin, and the good deal turned into the quarter's worst one.
That is the quiet thing about outcome based pricing in professional services. It does not just change how you charge. It moves delivery risk off the client and onto you, and most firms are taking that risk on with an estimating model that was built for the opposite world.
Outcome based pricing moves the estimate risk onto you
Start with who carries the cost of a wrong guess.
Under time and materials, the client does. You estimate a hundred days, it takes a hundred and fifty, the client pays for a hundred and fifty. Your margin per day is intact. The estimate was a forecast, and being wrong cost you a slightly awkward conversation, not money. So over years, estimating quietly became a sales discipline. The job of the number was to win the deal. Low enough to be competitive, high enough to look credible, and any gap between the estimate and reality was the client's to fund.
Outcome or fixed pricing inverts that. You name a price for a result, you sign, and now every day over the estimate is a day you eat. Estimate a hundred, deliver in a hundred and fifty, and the extra fifty are pure cost against a price that no longer moves. The estimate stops being a forecast you can revise on the invoice. It becomes the riskiest number in the business, because it is the number the whole deal's profit is now standing on.
This is the part the pricing conversation skips. Moving to outcome pricing is mostly a contracting exercise: a different clause, a milestone schedule, an acceptance definition. You can do it in a quarter. But the instrument you are using to set the price, the estimate, did not change when the contract did. It is still the sales number. You have changed who carries the risk without changing the thing that measures it.
The shift is real, and so is the downside it hands you
The market pressure here is not imaginary. The analysts and consultancies pushing the "PS 2.0" and Services-as-Software story are near-unanimous that services pricing is moving from time toward outcomes, value, and shared risk. Forrester argues AI delivery platforms squeeze the old hours-based margin and pushes firms toward value-based and shared-risk models. It also notes that management consultancies have already started putting fees at risk, with PwC reportedly doing it on around a third of engagements.
Worth holding the timeline honestly, though. The same research is early and contested. The pattern shows up clearly in maybe five to ten firms, not across the industry, and one credible read is that the billable hour is mutating rather than dying, with nobody having cracked the nut yet. So this is a transition, not a finished fact.
But notice what that does to the risk. A contested, gradual shift is the most dangerous kind to underwrite badly, because firms move into outcome deals one at a time, opportunistically, to win a specific account. The first one is rarely a considered change to how the business estimates. It is the same sheet, the same optimism, with the risk silently relocated. That is the trap: outcome pricing can lift your margins, or it can just hand you the client's downside, and which one you get depends entirely on whether your estimate can see the cost and its spread. Better selling does not save you here. The selling already worked, that is how you won the deal. Better estimating is the only thing that does.
A defensible outcome price needs a range, not a point
So what does an estimate built to carry risk actually look like. The first change is that it stops being a single number.
A sales estimate is a point. One figure, the one that wins. A risk estimate is a distribution: a likely cost, a plausible best case, and, the part that matters, a modelled downside. You are not pricing the hundred days you expect. You are pricing against the hundred and fifty you might hit, and deciding how much of that tail you are willing to absorb at the price on the table.
This is where the discipline gets concrete. If two senior people scope the same outcome and land twenty to forty percent apart, that spread is not noise to be averaged away. Under time and materials you could afford to ignore it, because the client funded whichever number turned out true. Under outcome pricing the spread is the risk. It is the literal range your margin lives or dies inside, and a price that ignores it is a price set blind. (This is its own problem with its own causes, which I have written about in why estimation variance is structural.)
The practical move is to estimate the variance, not just the point, and to price the outcome against the downside you can live with rather than the expected case you hope for. That is not pessimism. It is the difference between a price you can defend in a delivery review and one that only ever made sense in the sales meeting.
Contingency stops being padding and becomes priced risk
Most firms already have a crude version of this. It is called padding, and under time and materials it is faintly embarrassing, a hidden buffer the estimator slips in and hopes nobody questions, that the client ends up funding anyway.
Outcome pricing changes the status of that buffer entirely. It is no longer padding you hide. It is contingency you price, defend, and own, because there is no client behind it to absorb the miss. The buffer is now your risk capital, and capital you cannot see or explain is capital you will misprice.
So contingency has to come out of the estimator's head and into the model, attached to the parts of the scope that actually carry uncertainty. A clean integration into a system you have built ten times does not need the same buffer as the novel piece nobody has scoped before. Treating contingency as one flat percentage across the whole deal is how you end up overpricing the safe work, losing it to a competitor, and underpricing the risky work, winning it and bleeding on it.
Scope-change discipline gets stricter, not looser
The last piece is the one firms relax exactly when they should tighten it.
Under time and materials, scope creep is mildly annoying and quietly profitable. The client asks for the extra thing, you do the extra thing, you bill the extra hours. Change is just more billable work. So change control is loose by habit, because being loose made money.
Under outcome pricing that same looseness is a direct hit to margin. The price is fixed against a defined result. Every unbilled extra, every "while you're in there", every quiet expansion of what "done" means, is pure cost against a number that does not move. The agreed scope and the assumptions it rests on are the thing standing between your price and an open-ended commitment. Letting them drift without re-pricing is the most expensive habit a firm can carry into outcome work.
So the discipline inverts. The estimate has to record what it assumed, the scope has to be explicit enough that a change is visibly a change, and any movement against it has to trigger a deliberate decision about price, not an apologetic absorption. Tight acceptance criteria stop being paperwork and become the boundary of your risk.
What it takes to carry priced risk in a model
None of this needs a particular tool. It needs an estimate that holds a range instead of a point, contingency you can attribute to the risky scope rather than smear across all of it, and a record of what you assumed so a change is visible as a change. You can build that in a spreadsheet. It just tends not to survive contact with a real deal, because the next redraft quietly overwrites the assumptions and the buffer goes back to being invisible.
This is the part of Estii built for exactly that. Contingency is modelled as a range rather than a flat number, low, normal, and high, applied per item by its risk profile, so the novel work carries more buffer than the work you have done ten times, and the spread is in the model instead of the estimator's head. The rate card holds low and high margin thresholds and flags when a role's margin crosses them, so a price set too thin to absorb its own risk is visible while you are still negotiating, not after delivery. And every redraft snapshots a read-only deal version automatically, so the scope and assumptions a price was set against are recorded and a later change reads as a change you can re-price, not a quiet drift you find in the margin.
The broader version of this argument is that the billable hour is the unit of account for the whole business, not just the invoice, and repricing the top layer leaves the layers underneath broken. I have made that case separately in killing the billable hour breaks more than your invoice. This piece is the narrower one. Even if you grant all of that, the deal in front of you still needs a price, and the moment you sign an outcome you have made a bet on your own estimate.
Outcome pricing rewards the firm that can estimate the true cost and its spread. It punishes the one that can only estimate the number that wins. The shift moves the risk onto you whether or not your estimate is ready to carry it. The work is not learning to sell outcomes. The selling is the part you already know. The work is making the estimate honest enough to bet on.
