Nobody Has Priced This Yet
In my day-to-day, I speak to leaders at lots of consulting firms and we keep ending up in the same conversation: what do we do about AI and pricing?
The problem is not that there is no answer. The problem is that there are too many, and every one of them assumes something that may not be true: that the consulting firm can actually manage the cost of AI in the first place. That is a big assumption for an industry where, by micro-segment, plenty of firms are a few bad months from closing their doors.
Layer on the fact that clients want projects faster and cheaper while firms are trying to protect or grow margin, and you have a genuine standoff. AI does not resolve that tension. It just gives both sides a new variable to fight over.
The Fee Nobody Wants to Admit Is a Rate Cut
The first instinct I see is a technology fee, a separate line item for the AI tooling. I dislike it the most. It has the same feel as a baggage fee or a credit card surcharge: I am being asked to pay extra for something that should already be baked into the price.
To be fair, it is clean. It lets buyers compare rates and hours across vendors without the AI cost muddying the comparison, but it only works if nothing else moves. In practice, a firm adds the technology fee and then reduces hours to protect the client's price tag while padding margin. That is also, not coincidentally, one of the reasons so many consulting firms now push their own tools onto clients. A platform fee has the same effect with better optics.
Rates Are Not the Price of a Widget
The second instinct is to fold AI's benefit into the hourly rate and leave it there. This misreads what a rate actually is. Buyers tend to treat it like the price tag on a widget, but a rate is a compressed statement of market position, overhead, salary structure, and utilization targets. It is not a knob you turn to reflect one input.
Raise the rate and cut the hours and you can get to the same total price, but you now look more expensive than your peers on the number clients compare first. You also open two risk fronts at once. Internally, any hour overrun could make the project look unprofitable faster, because the cost rate is likely absorbing the AI cost. Externally, that same overrun lands harder on the client. Good luck getting that change order signed.
Fixed Price Is Where the Market Is Actually Heading, and Where the Sleight of Hand Lives
The market is pushing hard toward fixed price, and it is a harder deal for the firm to hold. You absorb the risk of the unknowns and, often, the client-caused delays. When it goes well, you keep the upside, which is the trade strategy consulting has run for years: set the price, then argue value, not rate.
Fixed price is also where AI-driven efficiency can flow straight to margin without touching the rate card at all, which is really a blend of the technology fee and the rate-cut tactics above, just with the visibility stripped out. Procurement departments know this, which is why large technology deals still demand a rate, role, and hours breakdown even inside a fixed price. They will tell you it is to spot where vendors read the scope differently. It is closer to reading tea leaves. You might catch something real, but you are more likely to draw the wrong conclusion and end up right back at a rate debate or a hidden hours cut anyway.
One dis-honorable mention I will not spend time defending: time and materials with a cap. It is the worst billing structure in the business, AI or not, and consultants should avoid it wherever they can. That is a different post.
The "AI as a Resource" Framing Is Seductive and Mostly Wrong
The idea I hear most often from firm leadership is treating AI as a headcount substitute, shifting hours from a human consultant to an AI resource the way you might move work from a junior to a senior. The work gets delivered faster, the hours-equivalent drops, and the effective rate starts to look like an offshore rate.
It is a tidy story until you ask how you actually bill that time inside a time and materials agreement. It is a bit like rating an EV against a gas car on an equivalent miles-per-gallon number. The comparison sort of works as a shorthand, and it mostly does not hold up once you look at what is actually being measured.
Outcome/Value-Based Pricing Is the Most Interesting Model That Struggles to Gain Traction
Outcome and Value-based pricing, where the firm takes a share of the upside it creates (or expects to create), is conceptually the cleanest answer to the AI question. You are pricing the benefit, not the cost, so the efficiency argument disappears entirely. It should be the model AI accelerates.
It has not gotten much traction, and I do not think AI changes that on its own. The model falls apart in cost centers like HR, where a lot of projects have no hard ROI to share in. And clients have real, structural resistance beyond the math: uncertainty about total cost, capitalization treatment, and the internal discomfort of potentially carrying an open-ended purchase order. Those are procurement and finance problems, not pricing-theory problems, and AI does not touch any of them.
What the Big Firms Are Actually Doing, and Why It Only Half Answers the Question
The public data point everyone cites is McKinsey now generating roughly a quarter of its global fees from outcomes-based arrangements, a real move away from the billable hour after decades of defining the industry around it. BCG has said AI-tied revenue is on track to roughly double as a share of the business between 2024 and 2026, and Bain has described AI and tech-enabled work climbing toward half its consulting revenue. That is the value-based and mixed-method paths, at scale, at firms with the balance sheet to absorb the transition risk while they figure it out.
What that data does not tell you is which of the seven models above is actually winning inside those numbers, or how a firm without MBB scale and MBB pricing power is supposed to run the same experiment. This has been building for a while. Trade press was already calling AI a threat to the billable-hour model in 2023, and the accounting and professional-services press is still framing this as an open transition, not a solved one, three years later.
The Model I Would Actually Build Toward
If I had to pick one, it is the mixed approach. Use AI-driven efficiency on things like meeting notes and document production as a built-in hours and duration reduction inside the existing rate, no separate line item, no drama. Reserve a fixed fee for the work where AI is doing the bulk of the lift: data conversion, integrations, heavy research, data analysis. That splits the two things buyers actually care about. Speed on the commodity work shows up as a lower total price. Value on the AI-heavy work gets priced on its own terms instead of getting buried in a rate argument.
In all cases, you will need to clearly separate how you calculate costs (which probably isn't going to change much) and how you calculate price (the numbers the client pays).
Even that is a starting hypothesis, not a house view. Whichever model a firm picks, the caution is the same: test it, validate the data, the assumptions, and the actual outcomes before any of it gets written into a contract. Give your teams time to adjust, because the early results will not match what you modeled. This has to become its own competency, folded into your methodology or accelerator function, with a real feedback loop measuring the impact on utilization and velocity as you go. Nobody gets this right by writing it into a master services agreement on faith.
The Question I Do Not Have an Answer To
I have walked through seven models and I still do not think there is a clean winner, because every one of them assumes a firm's cost structure and a client's procurement posture that vary wildly by micro-segment. I do not think anyone outside the MBB tier has actually solved this yet, and I am not sure MBB has either. They have just been able to afford the experiment longer than everyone else.
So I will leave this as the open question it actually is. If you are a buy-side leader reading a proposal built around one of these models right now, what would make you trust it? If you are running a consulting or IT services business trying to hold margin while AI compresses the work, which of these seven are you actually running, and is it working? I would genuinely like to hear how you are threading it, on either side of the table.
Comments
Post a Comment