There is a problem at the centre of every professional services firm's P&L right now, and the firms that survive the next five years are the ones that resolve it directly. The problem is that AI is compressing the time required to do the work, and the firms still selling time are watching their revenue base disappear into the compression.
The billable hour and what AI does to it
The mechanics are straightforward. A consulting engagement is priced at a daily rate or a fixed fee that translates back to a daily rate. The firm sells the engagement against the time it expects the engagement to take. Time is the substrate the price is built on. When AI compresses a task that used to take a four-person team a quarter into a two-person team a week, the firm faces a choice. Pad the engagement to defend the rate (which the customer eventually notices and stops paying). Walk away from the delivery work and retreat upmarket to strategy advisory (which is a smaller market than delivery and gets squeezed by the same compression on a longer time horizon). Both responses are visible in the Big-4 data already. PwC cut graduate hiring by 30 percent over three years. McKinsey eliminated five thousand support roles. Both are responses to a P&L that is no longer sustained by the old engagement count at the old rate.
Why billable-hour firms cannot adopt the technology
The deeper structural problem is that a firm whose entire P&L is the billable hour cannot adopt the technology that compresses the hour. Eudia named this directly in the legal context: incumbent law firms operating on a billable-hour model possess a structural disincentive to adopt technology that drastically reduces the time required to complete a task. The line generalises beyond legal. It applies equally to IT managed services firms whose P&L is engineer-hours, to accounting firms whose P&L is partner-hours, to consulting firms whose P&L is consultant-hours. Each of these categories has a structural disincentive to deploy the technology that would compress its own revenue base. The firms that adopt the technology cannibalise themselves; the firms that do not get displaced by ones that do. There is no comfortable position.
Outcome-based pricing as the way out
The way out, the only way out that has been demonstrated to work, is to stop selling time and start selling outcomes. Foundation Capital traced this evolution in their 2025 lessons-from-year-one piece as a four-step pricing spectrum: access-based, then usage-based, then workflow-based, then outcome-based. Each step up the ladder requires more confidence in the deliverable's measurable value, because each step is a stronger claim about what the customer is getting. AI delivery makes the highest rung of the ladder finally tractable, because the metering and measurement infrastructure that outcome-pricing requires is now buildable for the first time.
There is a related framing in the same Foundation Capital piece that is worth understanding. Seat-based SaaS pricing, the workhorse of the previous decade, is implicit insurance. The vendor charges a flat per-seat fee that absorbs the variance in how much each user actually consumes the product. As AI capabilities expand unpredictably, the cost of providing that insurance rises, because the variance gets larger. Outcome-pricing is the structure where vendor and customer have aligned exposure to that variance instead of the vendor absorbing it. The pricing model that survives the AI era is the one in which the vendor only gets paid when the customer captures the value, because that is the only structure where both sides can underwrite the unknown.
Operational proofs of outcome pricing across categories
The operational proofs of outcome pricing across categories are now visible in a way they were not eighteen months ago. Crosby in legal handles the $18B contract review market on volume-based pricing per contract, with a fifty-eight-minute median turnaround. Manifest OS in business immigration runs fixed-fee outcomes-based pricing and just raised a $60M Series A at a $750M valuation. Eudia, also in legal, took ARR from $2M to $20M in twelve months on an outcome-aligned model; their Knowledge Brain captures work the first time and applies it automatically thereafter, which is the technical substrate that makes the outcome measurement transparent enough for the customer to trust. Crescendo in call centres runs at gross margins four times the industry average, which is what outcome-aligned pricing on AI-delivered work produces when the operational substrate works.
The Big-4 firms know this. McKinsey reports 25 percent of projects already priced against outcomes. The shift inside the firms is real and accelerating. The question for an operator is whether to wait for the firms to finish restructuring, or to engage the firms that have already built around outcome-based pricing from inception.
Why customer skepticism of outcome pricing is rational
A reasonable counter to all of this is that outcome pricing is a marketing innovation that customers do not actually trust. The skepticism is real and rational; customers have been burned by outcome-pricing schemes in software for fifteen years, where the "outcome" was defined by the vendor in a way that made it almost impossible not to charge. The way to address the skepticism is the way Eudia did. Build the underlying measurement infrastructure first. Capture the baseline transparently before the engagement starts. Meter the delivered work against the baseline so the customer can see exactly what they are paying for. The customer's trust is earned by the transparency of the metering, not by the headline pricing structure. Outcome pricing without measurement is a sales gimmick. Outcome pricing with measurement is the only pricing structure that survives the era.
There is one more thing worth saying about the unbillable hour, and it is the point that Hammer made in 1990 in different language. The principle that "organise around outcomes, not tasks" is the first of his seven reengineering principles. Hammer was not writing about pricing in 1990; he was writing about workflow design. But the principle is the same one. Tasks are inputs. Outcomes are outputs. The firms that organise around tasks (and price around tasks, and measure around tasks) are systematically misaligned with what the customer is actually buying, which is the outcome. Hammer named this thirty-five years ago. The firms that have spent thirty-five years not internalising the lesson are now finding that the technology forces the conversation in a way that the workflow-design discipline never quite did.
Outcome-based pricing is not a marketing innovation. It is the only pricing structure that survives software economics meeting professional services. The firms still selling time will spend the rest of the decade defending a rate the AI is making indefensible.