Hourly billing is a moral hazard now.

The incentive problem, stated plainly.

If I bill you for the hours I spend, and Claude compresses those hours, I am penalized for using Claude.

That is the whole essay in one sentence. Everything else is elaboration.

Consulting firms that bill hourly have an incentive structure that rewards slow work. This is not a new complaint. Clients have known it, written about it, gone in circles about it for thirty years. But it was tolerable for a long time because the slowness was load-bearing. A senior consultant's hour was the market price of a senior consultant's judgement, and the judgement took the hour. Billing for the hour was billing for the judgement. The two were coupled.

Claude uncoupled them. A senior consultant's judgement is now applied to a Claude-drafted artifact at roughly one-third the time cost of producing that artifact from scratch. The judgement is still senior. The hour count collapsed. Any hourly-billed firm now has to choose between charging less (which destroys the partner economics they built on) and charging the same (which means they keep the productivity gains and the client funds the AI tool through the same line item that funded the junior consultant it replaced).

Neither choice is ethical. The first is a business suicide. The second is the moral hazard.

Why hourly billing made sense for a century.

Before we reform something, we should understand why it existed.

The hourly billing model, as applied to professional services, is roughly a century old. It emerged in law and accounting and was adapted to management consulting by the mid-twentieth century. The logic was simple: professional work is labor-intensive, labor quality varies, labor cost varies, and the only way to track what a client is paying for is to measure the labor.

For most of that century, the measurement was approximately honest. A senior lawyer who billed 2000 hours in a year was working approximately 2000 hours on things a client would consider valuable. The hour was a proxy for the attention. The attention was the product.

The model also had a property that was considered a feature at the time. It made the vendor's cost structure legible. The client could see where the hours were going. The client could question a bill. The vendor had to justify the work. A fixed-fee engagement gave the vendor an incentive to cut corners; an hourly engagement gave the vendor an incentive to do the work, and the check on overbilling was the partner's reputation and the client's right to audit.

This is the background the consulting industry inherits. It is defensible for the century in which it developed. It is not obviously defensible any longer, because the underlying condition - that attention and hours are correlated - stopped being true for cognitive work as of roughly 2024.¹

What Claude broke about the model.

The link between attention and hours broke specifically, and measurably, for a set of deliverables.

3.1 Document production.

A ninety-page consulting report used to take three weeks of senior writing time. It now takes three days of senior editing time plus about eighty dollars in API cost. The attention is still senior. The hours are not the same.

3.2 Code review and audit.

A full code audit of a mid-sized SaaS used to take two weeks. The senior now uses Claude to generate an initial finding set in under an hour, then spends three days validating, contextualizing, and deepening the findings. Same level of audit. Different hour count.

3.3 Research synthesis.

A market landscape used to take a junior ten days of desk research. It now takes the senior two days of validated prompting and editing. The junior is not out of the loop - they still verify every citation - but the core synthesis labour compressed by a factor of four or five.

3.4 Artifact generation.

Sprint plans, decision matrices, roadmaps, pricing tables, competitive analyses - the generic consulting output - used to be the junior's job. The junior spent days on each. They are now twenty minutes of prompting plus thirty minutes of senior review. The senior does not charge the client any differently.

These are not marginal gains. They are order-of-magnitude changes in the hours required to produce the work. A firm that bills hourly and has absorbed these tools has two options: keep the rate the same and collect the compression as margin, or lower the rate and protect the client. The first is what most firms are doing. It works until the client figures it out.

The math of fixed-scope, worked out.

Here is how the numbers move when the pricing model moves.

Imagine a standard six-week security engagement: scoping, reconnaissance, testing, reporting. Under the hourly model at $350 per hour, six weeks of a senior tester plus one week of a senior writer comes out to approximately 280 hours, or $98,000. The deliverable is a report.

Now introduce Claude in the documentation phase. The writing week compresses to two days of senior editing. The testing weeks compress by about 25% because the finding-writeup phase happens in parallel with testing rather than after it. The total engagement labour drops to about 175 hours. At the same $350 rate, the hourly firm would have to invoice $61,250. That is $36,750 less revenue for the same deliverable. The partnership cannot support the same compensation at this level.

The hourly firm's options: raise the rate to $560 per hour (embarrassing), hide the compression (dishonest), or keep billing at the old hour-count (fraud). All three are being practiced right now. All three eventually fail.

The fixed-scope alternative: the engagement is $65,000, delivered in six weeks. The firm gets to keep the productivity gains as margin. The client pays a defensible number that reflects the value of the deliverable, not the cost of labor that no longer correlates to attention. The client cannot be overcharged for time, because the contract is not about time.

Look at the third row. That is the row where the client pays full legacy price for a deliverable that took the firm 40% less labor, and the firm keeps the delta as pure margin. If you are a client doing procurement for a hourly-billed consulting engagement in 2026, this is the row you should be worried about. It is the row most hourly firms would prefer you did not study.

Objections from inside the consulting industry.

5.1 "Fixed-scope encourages corner-cutting."

It does. The check on corner-cutting is the published scope document. We publish scope with deliverables named by artifact. "90-page handbook with A, B, C chapters, N+ findings, evidence zip, quarterly refresh." If the firm delivers less, the client does not pay. The scope document is the contract, not the hour ledger.

5.2 "Clients do not know what they need, so fixed-scope is dishonest."

This is a real problem and the answer is a paid discovery engagement. A two-week scoping sprint that ends in a proposal with a fixed price. The client pays for the scoping at a modest, fixed number - two to eight thousand dollars depending on complexity - and the larger engagement price is then anchored in a shared understanding. If the client declines the larger engagement, they at least leave with a planning document that was worth the scoping fee.

5.3 "You can't fixed-scope creative work."

You can fixed-scope the container around creative work. "Fifteen long-form essays, each between 2,000 and 4,000 words, in our published voice, first draft within two weeks, two revision cycles included, published with schema markup." The shape is fixed. The creativity lives inside the shape. This is how book publishing has operated for three centuries; it is a solved problem.

5.4 "The hourly model lets clients audit our work."

In theory. In practice, clients rarely audit hours in detail. The bills that arrive are lump sums with vague line items. What actually gives clients confidence is the published deliverable, not the time ledger. Fixed-scope shifts the audit from "did you spend the hours you said" to "did you deliver the thing you said." That is a better audit.

5.5 "Our partners get paid by billable hours."

This is a compensation design choice, not an economic law. Partners can be paid on fixed-fee engagement margin. Several large firms have moved in this direction quietly over the past two years. The partners who resist are the partners whose relationship to their own firm is structured around the hourly accumulation. That is a partnership-design question, not a client-pricing question.

What fixed-scope does not fix.

Fixed-scope pricing is not a panacea. It has real failure modes and the honest way to argue for it is to name them.

Fixed-scope transfers the scope-creep risk from client to vendor. Under hourly, the client was on the hook for their own scope expansion - if the project grew, the hours grew, the bill grew. Under fixed-scope, the vendor has to hold the line on scope or eat the overrun. A vendor that does not hold the line on scope will lose money. A vendor that holds the line too aggressively will look inflexible.

The fix is the change-order mechanism. Every fixed-scope contract has a named path for scope changes: if the client requests additional work, the vendor prices the additional work, the client approves or declines, and the additional work becomes its own fixed fee. We have a published change-order template on our rules of engagement page. The template is three pages. It is the load-bearing instrument of fixed-scope delivery.

Fixed-scope also requires the vendor to be a good estimator. A firm that consistently under-estimates fixed scopes will go bankrupt. A firm that consistently over-estimates will lose bids to firms with better estimation. This puts a premium on operational maturity. The smaller or newer the firm, the more dangerous fixed-scope can be. Our answer: we started with a narrow range of engagement shapes - audit, handbook-driven engagement, retainer - and we hold our scope definitions tight. The firms that get hurt by fixed-scope are the ones that accept any scope someone offers and discover the cost later.

The retainer, reconsidered.

The retainer is the least-discussed of the three pricing shapes and the one that most benefits from the AI compression.

A monthly retainer gives the client access to a defined amount of senior attention over a defined period. In the pre-Claude world, a $10,000-per-month retainer bought the client roughly thirty senior hours. After Claude, the same $10,000 per month buys the client access to roughly the same amount of deliverable output, but the senior hours involved are closer to twelve. The retainer does not need to be repriced downward. It needs to be repriced upward in deliverable capacity.

What clients now get from our monthly retainers: a monthly handbook refresh, an editorial program producing eight to twelve artifacts, an evaluation harness running continuously, a shared dashboard. Under the old model, none of this would have fit inside the retainer. Under the new model, all of it fits.

The retainer also solves a real problem that fixed-scope does not. Ongoing work - security monitoring, content publishing, experimentation - is inherently open-ended. You cannot fixed-scope a year of monitoring. You can retainer it, at a defined monthly fee, with defined deliverables per month, with a defined cancellation term. This is where ongoing value lives.

Our own mix is 40% engagement revenue, 45% retainer revenue, 15% licensing. The retainer is the largest line item and growing. The hourly line item is zero, and will remain zero.

Closing. The contract shapes the product.

The shape of the contract shapes the work. This is the oldest observation in all of consulting.

An hourly contract shapes the work toward explicable hours. The consultant takes longer than necessary because the client is paying for the hour. The client demands hour-by-hour timesheets because the bill is denominated in time. The document that arrives at the end is shaped by the time-tracking discipline that produced it, not by the substance of the engagement. This is not a moral failing. It is a structural consequence of the contract.

A fixed-scope contract shapes the work toward explicable deliverables. The consultant writes as fast as they write well, because the client is paying for the artifact. The client examines the artifact, not the time ledger. The document is shaped by what needs to be in the document, not by how long the document took to produce. When the AI produces the document at ten percent of the old cost, the artifact is the same; the contract is the same; only the margin shifts, and the margin is the firm's problem, not the client's.

The moral hazard of hourly billing under AI is real. It is not a rhetorical complaint. It is a structural misalignment between what the client is paying for and what the vendor is producing. The only way to resolve it is to move the contract to a different basis.

We billed fixed-scope from day one because we were starting the firm after the shift had already happened. Older firms will have a harder time. They will do it eventually, because the alternative is being outcompeted by firms that offer the same deliverable at a lower number. The market does not tolerate the hidden-compression posture forever. Clients are getting smarter every quarter.

If you are commissioning consulting work in 2026, here is our advice. Ask for fixed-scope. Ask what the deliverable is, in specific detail. Ask what changes to the deliverable trigger a change order. Do not ask how many hours. Hours are the wrong unit now, and any firm that still measures in them is a firm whose contract is shaping its work in the wrong direction.

1. The shift is gradual and uneven across task types. Writing and editing compressed earlier and more completely than, say, novel exploit research. We wrote about this in "Claude is not a pentester" and will return to it in a forthcoming essay on consulting-firm P&L. Back to text