AI pricing in professional services: Summary & key takeaways
The data: Firms running complex AI agents or an AI-first model report achieved consultant day rates of $1,364, compared with $962 for firms using AI only as a chat assistant or not at all (SPI Research, 2026 Impact of AI on Professional Services).
The challenge: When fees are calculated from billable hours, AI-driven efficiency can reduce revenue even when hourly rates stay unchanged.
The opportunity: Price the outcome and deliberately decide how much efficiency to retain or share.
By model: Move predictable hourly work toward deliverable pricing, protect the economics of fixed-fee work, and define retainers around monthly outcomes.
The foundation: Understand actual delivery costs and achieved rates to make better pricing decisions.
Ask a room of services leaders what AI will do to their pricing and most say the same thing: discounts. Fewer hours, smaller invoices, clients wanting their share. I get the instinct. I'm a chartered accountant by training, and when hours fall and the rate doesn't move, the math is hard to argue with.
The data says something different. SPI Research's 2026 Impact of AI on Professional Services study found firms running complex agents or an AI-first model report achieved consultant day rates of $1,364, compared with $962 for firms using AI only as a chat assistant or not at all.
The research doesn't establish how much of that difference comes from pricing, delivery efficiency or the mix of work. For illustration, a $10,000 engagement delivered in ten days generates $1,000 per day. Deliver the same engagement in six days and that rises to roughly $1,670 without charging the client more. That isn't necessarily how SPI calculates its benchmark, but it shows why lower delivery effort doesn't automatically mean lower commercial returns.
So where does the pressure on fees really come from, and what can firms do differently? It takes three moves: understand why hours-based pricing can reduce revenue, price the outcome with a clear understanding of delivery costs, and check every price against what the work actually cost.
Mossy covered where the saved hours go. This piece is the commercial side: what you charge for the work in the first place.
Smaller invoices can come from the formula, not just AI
A lot of services pricing is cost-plus in disguise. Estimate the hours, multiply by the rate, add contingency. Under that formula, AI reduces the hours required, so the total fee falls even if the hourly rate stays unchanged. That's not technically a discount. It's a consequence of how the work is priced.
Clients are making the same calculation. 59.9% of firms say clients now expect AI-enabled delivery, up from 45.0% in 2025. Expecting AI isn't the same as expecting a lower price. But if your fee is a function of hours, "how many fewer hours?" is the question you've left the client to ask. One customer recently told our customer success team that their clients have started asking for an "AI dividend": a discount, because AI makes the work faster.
The bigger problem is when you find out. As an ex-financial controller, this is the part that worries me most. In many services businesses, the real cost of an engagement only shows up after it ends. By then the price is long agreed, and margin is already won or lost. If you set prices from estimated hours and only learn the actual cost afterward, you can't reliably assess whether the price was right.
The opportunity isn't necessarily to hold every price. Sometimes sharing AI savings with clients makes commercial sense, particularly if it helps win more work. But that should be a deliberate decision, not an automatic consequence of estimating fewer hours. The question is how much of the value created by AI you retain, and how much you pass on.
)
Price the outcome, and understand the cost of delivery
Cost helps you understand the economics of an engagement. Value helps determine what you can charge. The principle is the same across all three pricing models, but the pressure lands in different places.
Hourly or T&M: the saving shows up on the invoice. When AI reduces billable hours, revenue falls unless the pricing approach changes. For predictable, AI-heavy phases like research and first drafts, consider a fixed fee per deliverable rather than billing for time. Keep T&M for work where scope or effort is genuinely uncertain.
Fixed fee: the saving shows up in the margin. Price the deliverable based on its value to the client, not just your revised estimate. If AI reduces delivery effort while the fee remains unchanged, the firm can retain the efficiency through improved margin and capacity for additional work. The next quote is where you decide how much of that value to retain or share.
Retainer: the saving surfaces at renewal. "You get 60 hours a month" invites the question of why it's still 60 once the client knows AI is involved. Define monthly outcomes instead (campaigns launched, reports delivered, response times), tier them, and price the AI-enabled tier above the basic one where the additional value justifies it. Understand what each tier costs to deliver.
Three models, one move: describe what the client gets, not just the hours, and understand the cost of delivering it. Another customer told us they're moving exactly this way. Clients pay for the deliverables, and how long the work takes stops mattering to the fee.
For agencies, the retainer is the urgent one. The same logic covers fixed-fee campaigns and client projects: price what the work is meant to achieve, not the creative hours behind it. For consultancies, price the engagement on the deliverable and the decision it supports.
The step most firms skip: checking the price against the cost
Coming from finance, this is the part I find most important. Changing your pricing model is one thing, but without understanding what the work actually costs, you're still guessing.
Once you price by deliverable, your effective day rate becomes an output rather than an input. You need to understand what you earned relative to the effort invested, what it cost to deliver, and whether margins improved. AI makes historical estimates less reliable, so learning from actual delivery becomes even more important.
SPI's data shows the same gap. Rate analysis scores 1.97, among the weakest areas it measured, and 45.5% of firms say AI hasn't helped with it at all. AI can't price your work from actuals nobody has captured cleanly.
After each engagement, ask three things. Did the estimate hold? Did the margin hold at the price you charged? And would the client have paid more for what they got?
Most firms investigate why projects go over budget. Far fewer systematically ask whether they underpriced the work that went well.
On the sales calls I join, the CFO is increasingly in the room, and the concern is rarely the tool. It's the risk to financial operations and reporting. But the finance stack doesn't change: GL, AP/AR, tax and audit. What changes is everything upstream that decides margin: estimates, capacity, time, cost and billing decisions. Pricing sits on top of those numbers. If they arrive late, or not at all, every price is a guess.
That's what we're building Teamwork.com to do. As the agentic PSA, it connects projects, resources and financials, so you can see margin while the work is still running, not after the engagement ends. That's the feedback loop. And when you come to price the next piece of work, you can quote from what comparable work actually cost, instead of from a guess.
Where to start this quarter
You don't need a firm-wide pricing reset. Take one service line and run the whole loop once, in four steps.
Pick the line where AI has changed delivery the most. That's where your current price is most likely to be out of date.
Work out what a deliverable costs now. Include people and AI. That gives you a baseline for understanding your margins.
Write down what the client gets. Faster, better, less risk? If you can't say it in a sentence, you can't price it. Blair Enns, founder of Win Without Pitching, suggests going straight to the client with three questions: what does success look like, how will we measure it, and what is it worth to you? Their answer beats your guess.
Offer a priced alternative at the next proposal or renewal. Put a fixed-fee or tiered option next to the hourly one and see which clients choose. Blair goes a step further and recommends offering three options, led by the most ambitious, because clients tend to retreat from the extremes and settle in the middle (he covers both ideas in our webinar with him).
Then close the loop. When the engagement ends, put the price next to the actual cost and ask the three questions again.
AI is changing what it costs to deliver professional services, but it doesn't automatically determine what clients should pay. Firms can retain the efficiency, share it with customers or use it to create more valuable services. Who captures that value is a commercial decision.
)
)
)
)
)
)
)
)
)
)