The Repricing Playbook: What to Do When Clients Refuse to Pay Hourly for AI-Assisted Work
· By Practical AI Institute · Professional Services
When clients refuse to pay hourly for AI-assisted work, firms have three viable moves: fixed-fee repricing of commoditized deliverables, outcome-based pricing on high-judgment work, and an AI-disclosure policy that protects trust. Firms that reprice within two quarters protect margin; firms that wait absorb 15–30% realization compression.
This playbook is written for managing partners and CEOs of professional services firms in the $5M–$100M revenue range — law, accounting, consulting, and engineering — whose clients have started redlining invoices. It covers the client-side billing-guideline shift, the cost of waiting, the three repricing moves in sequence, and who should own the project internally.
Why are clients refusing to pay hourly for AI-assisted work in 2026?
Because their own procurement and legal-ops teams now run AI-assisted invoice review and have rewritten billing guidelines accordingly. Meta updated its outside counsel guidelines to flag and reject line items AI could have performed. Zscaler's published guidelines bar passing AI-generated work costs to the company. UBS added AI-specific billing provisions in early 2026.
The pattern extends well beyond legal. A 2026 General Assembly survey of 258 director-and-above leaders at consulting, accounting, and legal firms found that 79% say AI is changing pricing conversations, 42% report clients actively questioning their pricing model, and only 37% are addressing it proactively. That last number is the strategic opening: roughly two-thirds of your competitors are still reacting instead of repricing.
| Buyer | Guideline change | What it means for firms |
|---|---|---|
| Meta (outside counsel) | Flags and may refuse payment for tasks AI could perform (summaries, routine drafting, settled research) | Commodity-layer hours come back redlined |
| Zscaler | Time and cost of AI-generated work product cannot be billed to the client | Hourly billing for AI output is contractually dead |
| UBS | Added AI-specific provisions to billing guidelines in early 2026 | Financial services buyers are formalizing the same posture |
| Mid-market buyers (your clients) | Adopting the same language 12–18 months behind enterprise buyers | The repricing window is open now, not indefinitely |
Note the procurement subtext: these guidelines arrive alongside vendor questionnaires about SOC2 attestation, data residency, and AI governance. Clients are not only refusing to pay for AI-assisted hours — they are auditing whether your firm uses AI responsibly on their data. Your repricing response and your AI-governance posture get evaluated together.
What does it cost a professional services firm to delay repricing?
Practical AI Institute's analysis projects 15–30% realization compression for firms that keep hourly billing while clients enforce AI guidelines. On a $20M firm at the Rosenberg MAP benchmark of roughly 86% realization, a 15-point slide to 71% removes about $3M in collected revenue with zero reduction in workload.
The mechanism is asymmetric and quiet. Realization erodes invoice by invoice — a redlined research memo here, a rejected document-review block there — so no single quarter forces a decision. Meanwhile, Thomson Reuters' 2026 State of the US Legal Market reports technology spending rose 9.7% in 2025: firms are paying for the efficiency tools and giving the efficiency away. Worse, 52% of corporate counsel plan to insource more work within five years, meaning the hours you fail to reprice today may not exist to reprice tomorrow.
There is a counterintuitive grace period. Over 80% of senior corporate counsel do not yet require their outside firms to use AI. The demand side is enforcing AI economics before mandating AI adoption — which means a firm that moves within two quarters can reprice on its own terms rather than under a client ultimatum.
What are the three repricing moves, and which work gets which model?
Segment every deliverable by judgment intensity, then price each tier differently: fixed fees for commoditized output AI accelerates, outcome-based pricing for high-judgment advisory work, and a written AI-disclosure policy covering both. Accounting firms running this structural repricing report rate increases of 25–30% on AI-assisted work moved to value and fixed-fee models.
| Move | Applies to | Pricing mechanism | Margin effect |
|---|---|---|---|
| 1. Fixed-fee repricing | Commoditized deliverables: routine drafting, first-pass research, standard filings, recurring compliance work | Scoped flat fee per deliverable or phase; AI efficiency accrues to the firm | Efficiency gains convert to margin instead of disappearing as unbilled hours |
| 2. Outcome-based pricing | High-judgment work: negotiations, strategy, expert testimony, deal structuring, advisory | Fee tied to defined outcomes, milestones, or value bands — never to hours | Decouples revenue from time entirely; immune to AI invoice review |
| 3. AI-disclosure policy | All client work | Written policy in the engagement letter: where AI is used, how data is protected, how pricing reflects it | Protects trust and removes the discount conversation before it starts |
Move 1: Fixed-fee the commodity layer first
Start with your highest-volume, most repeatable deliverables — the exact line items client guidelines target. Run a Process Teardown Protocol on each: document current hours, apply AI-assisted workflow time, and set the fixed fee at or slightly below the historical hourly total. The client sees price certainty and a number no invoice reviewer can redline; the firm keeps the efficiency delta. Legal industry data shows firms billing flat fees collect nearly twice as fast as hourly peers, and 71% of clients prefer flat fees for entire matters.
Move 2: Outcome-base the judgment layer
High-judgment work was never really priced by the hour — the hour was a proxy for value that clients tolerated. AI removed the tolerance, not the value. Define outcomes in the engagement letter (deal closed, dispute resolved within a band, audit completed by date, regulatory approval secured) and attach fees to those outcomes. KPMG, PwC, and RSM are publicly rebuilding pricing around value and outcomes for exactly this reason; mid-market firms can move faster than any of them.
Move 3: Publish the AI-disclosure policy
A disclosure policy converts AI from a billing liability into a trust asset. The minimum viable policy answers four questions in writing:
- Where is AI used?
- Named task categories (summarization, first-pass drafting, document review), with human review checkpoints stated explicitly.
- How is client data protected?
- Enterprise AI tiers with data-use opt-outs, SOC2-attested vendors, data residency commitments, and HIPAA safeguards where healthcare client data is in scope.
- How does pricing reflect AI?
- Fixed and outcome-based fees already price in AI efficiency — clients are never billed hourly for machine-generated output.
- Who governs it?
- A named accountable executive and an AI governance standard reviewed quarterly.
Token and compute costs belong in your cost model, not on the client invoice. A firm spending $2,000–$60,000 per year on AI tooling and recovering it through repriced fixed fees is structurally healthier than one itemizing "AI usage" as a billable line — which the new guidelines explicitly reject.
How fast can a firm execute the repricing — and in what order?
A focused firm can complete the full sequence in 90 days: weeks 1–3 for a Revenue-Drain Audit of realization leakage by client and matter type, weeks 4–8 for Process Teardown and fixed-fee construction on the top ten deliverables, weeks 9–13 for outcome-based pilots, the disclosure policy, and partner enablement.
The sequence matters because each move funds the next. The Revenue-Drain Audit quantifies exactly which clients and matter types are compressing realization — firms under $2M in fees benchmark at 92.5% realization and firms over $20M at 86.2%, so you are measuring your slide against a known baseline. Fixed-fee conversion then stabilizes the commodity layer before outcome-based pilots touch your most valuable relationships. A Business Impact Dashboard tracks realization, effective rate, and margin per matter weekly, so partners see the repricing working inside one quarter — an ROI timeline short enough to hold partner consensus together.
Who should lead repricing — a full-time Chief AI Officer or a fractional CAIO?
For firms between $5M and $100M, a fractional Chief AI Officer (fCAIO) at $15K–$40K per month is the economically rational owner: 35–65% lower executive OPEX than a $520K–$910K all-in full-time CAIO, with a deployed roadmap in 90 days — the same window the repricing itself requires.
Repricing is not a pricing-committee memo; it is an AI transformation program with a pricing output. Someone must rebuild workflows around AI, write the governance policy procurement teams will audit, model the fee economics, and survive partner pushback. That is executive work, but it is not permanent executive work — which is precisely the fractional case.
| Factor | Fractional CAIO | Full-time CAIO |
|---|---|---|
| Annual cost | $180K–$480K ($15K–$40K/month) | $520K–$910K all-in ($450K+ base) |
| Time to deployed roadmap | 90 days | 6–12 months including search and ramp |
| Executive OPEX impact | 35–65% reduction vs. full-time | Full C-suite load on a $5M–$100M P&L |
| Fit for repricing mandate | Built for a defined 90-day transformation sprint | Justified only if AI leadership is a permanent standalone role |
The math is unforgiving in the right direction: a $20M firm facing $3M of realization compression can fund a full year of fCAIO leadership for roughly 6–16% of the revenue at risk.
Frequently asked questions about repricing AI-assisted work
- Should a professional services firm disclose AI use to clients?
- Yes — proactively and in writing. Client billing guidelines and procurement questionnaires increasingly assume AI use; firms that disclose first, with SOC2, data residency, and governance commitments attached, convert the question into a trust advantage. Firms caught using AI silently under hourly billing face fee disputes and credibility damage.
- Do we have to cut prices because AI made the work faster?
- No. You have to stop selling time. Fixed fees set at or near historical totals give clients price certainty while the firm retains the efficiency gain; accounting firms executing this structural repricing report 25–30% higher effective rates on AI-assisted work, not discounts.
- How long does a firm-wide repricing program take?
- 90 days for the core sequence: a Revenue-Drain Audit in weeks 1–3, fixed-fee conversion of the top ten commodity deliverables in weeks 4–8, and outcome-based pilots plus a published AI-disclosure policy in weeks 9–13, tracked weekly on a Business Impact Dashboard.
Reprice before your clients do it for you
Every quarter of delay is realization you do not get back. Practical AI Institute's fractional Chief AI Officer engagement ($15K–$40K/month) delivers the full repricing sequence — Revenue-Drain Audit, Process Teardown Protocol, fixed-fee and outcome-based fee architecture, AI-disclosure policy, and a live Business Impact Dashboard — on the 90-Day AI Transformation Roadmap.