The billable hour has been pronounced dead so many times that the obituaries form their own genre. The value pricing movement has been writing them since the 1980s, bar associations join in every decade or so, and every few years a major firm announces a bold move to fixed fees that quietly reverts within two budget cycles. Through it all, the six-minute increment kept its grip on consulting, law, accounting, and agencies. I think the current obituary is the real one, not because the critics finally won the argument, but because the thing the hour was measuring has stopped existing.
My vantage point: thirteen years on the commercial side of enterprise software, at companies like Aiven and Ververica, selling subscriptions next to professional services teams that price their work in person-days, plus years on the buying side as a marketing leader signing agency retainers, watching invoices arrive itemized by the hour and being unable to tell which of those hours had produced anything.
The hour was never the product
The billable hour was always a proxy with two jobs. The first was approximation: when professional output was produced by humans reading and drafting, effort correlated with value closely enough that pricing time was a tolerable way of pricing work. A twenty-hour contract review generally contained more diligence than a two-hour one. The correlation was never perfect, but it was real, and it let both sides agree on a number without defining what the work was actually worth.
The second job was trust. A client who cannot evaluate the quality of a tax structure can still audit a timesheet. The hour was a verification mechanism for invisible work: you could not see the thinking, but you could count the time, and counting the time felt like oversight. Entire procurement departments are built on that feeling.
Why it survived every previous obituary
The case against the hour is old and has always been correct: it rewards slowness, sets the firm's incentive against the client's, and measures input in a business where the client only cares about output. The American Bar Association convened a commission that said all of this in 2002. Ron Baker has spent decades building the replacement playbook. Yale Law School hands its students a primer on what billing 2,200 hours a year actually does to a day, and Clio's Legal Trends data shows only about a third of a lawyer's workday ends up billable at all. None of it killed the hour.
It survived because every alternative required the negotiation everyone wanted to avoid: agreeing in advance on what the work was worth. Fixed fees need precise scope, and professional work resists precise scoping. Outcome pricing needs an agreed definition of the outcome, which in many engagements is harder than the work itself. The hour let both sides skip all of that. It was not a good system; it was the cheapest available truce, and as long as time and value stayed correlated, the truce held.
Agents broke the correlation
What changed is not the argument but the production function. The research memo, the first contract draft, the campaign concepts, the financial model: this was the billable middle of professional services, where most of the hours lived. Goldman Sachs estimated that 44 percent of legal work tasks could be automated with generative AI, and an agent already produces credible first versions of all of it in minutes. The professional's job shifts to reviewing the output and taking responsibility for it. The deliverable that defensibly cost forty hours in 2023 now costs three, and the three are a different kind of hour: senior judgment applied to machine output, not junior time spent producing it.
Once the same deliverable can cost forty hours or three depending on tooling, the hour stops carrying information. Bill honestly and revenue falls in proportion to your own efficiency. Keep billing forty and you are misrepresenting how the work was done, a position with a short shelf life, because clients have access to the same models and have often tried the task themselves before picking up the phone. The information asymmetry that made time auditing feel like oversight has collapsed from both ends at once.
The pyramid was the business model
The hour was not just a pricing unit; it was the load-bearing wall of how firms are structured. The leverage model, a few partners selling work that armies of associates and analysts execute, works because the firm bills the junior at a multiple of their salary. Agents eat precisely that layer: the work that was profitable enough to bill out at multiples is the work machines now do best. What remains is the partner's judgment, which never scaled, and which the pyramid existed to amplify.
There is a second-order problem I find more worrying than the pricing. Juniors became seniors by doing the grunt work; a decade of document review and model building is how judgment was trained. If agents take the grunt work, firms keep the seniors they have and lose the mechanism that produced them. Thomson Reuters' Future of Professionals surveys are full of optimism about the hours AI frees up; almost nobody is asking what those hours were quietly teaching. The industry will run for a decade on its existing stock of judgment and then find the pipeline behind it empty.
What gets priced instead
Strip away the hours and what clients were buying underneath is accountability: someone who knows which of the plausible outputs is the right one, will sign their name to the filing, and answers the phone when it goes wrong. Output was always abundant relative to judgment; the cost of production just hid that, the same way I argued in the Sulci essay that the model itself was never the moat. When production gets cheap, the price tag migrates to whatever cannot be generated, and in professional services that is the willingness to be responsible for the result.
The models that fit this reality already exist at the edges: fixed fees for defined deliverables, with the firm keeping the efficiency gain as margin; success fees where the outcome is measurable; subscriptions that buy access to judgment rather than hours. Software made this exact transition long ago. Nobody prices software by the engineering hours it took to write, and professional services could resist that logic only while their marginal cost of production stayed high. It just stopped being high.
The counterargument worth taking seriously
A residue of work will rightly stay hourly, because presence genuinely is the product: a day in court, a workshop facilitated, an on-call rotation staffed. Cost-plus government contracting will keep timesheets alive through sheer procedural inertia. And the trust function still needs a replacement, because fixed fees and outcome pricing have their own pathologies: scope disputes, sandbagged definitions of success, vendors doing the minimum that satisfies the letter of the deliverable. The hour was a bad answer to a real question, and the question, how you buy work you cannot evaluate yourself, does not disappear with it. My expectation is not a clean death but an inversion: the hour becomes the exception that requires justification, the way fixed fees used to be.
Where this leaves us
If you sell time, write down what your clients would still pay for if they could generate your deliverables themselves, because increasingly they can. Whatever survives that exercise, the judgment and the willingness to own the outcome, is the actual product, and it deserves a price that has nothing to do with a clock.
If you buy time, stop. Ask what the result costs, who is accountable for it, and what happens if it is wrong. The hour survived forty years of obituaries because both sides found it convenient to keep measuring the wrong thing. The measurement just lost its meaning, and there is nothing convenient left to save.