# The Efficiency Trap: Why AI Is Forcing Services Firms to Rethink Pricing

> AI makes delivery faster, which quietly shrinks the invoice under hourly billing. How we rebuilt pricing around the deliverable, not the hours, and the catch nobody puts on the brochure.

- Source: https://verygood.ventures/blog/ai-services-pricing/
- Published: 2026-09-03
- Author: Andrew King
- Tags: AI, Business Value, Enterprise, VGV

---

For most of professional services history, getting better at the work meant earning more. AI breaks that. If you bill by the hour and a tool lets your team deliver the same result in less time, you have cut your own invoice. The thing that makes you faster makes you poorer. That's the trap. But it is not the destiny.

That bind is one every services firm is walking into, and most are not saying it out loud. We decided to work through it in the open, starting with our own pricing model. What follows is the thinking from Erik Manley, CFO, and Andrew King, Director of Solutions Architecture, who rebuilt how we price, including the parts they have not figured out yet.

## The trap is built into the billing model

Start with the mechanics. Under time-and-materials, the client pays for hours and expects a certain output. Introduce AI, and the expectation shifts fast.

"Clients are going to expect us to be using AI, and they're going to expect more output for the same number of hours, or fewer hours for the same output," Erik said. "All that really means is they're going to be looking to pay a lower price for the same output."

Play that forward across a market of firms all competing on hours, and there is only one direction it goes. "T&M is just going to end up being a race to the bottom," Erik said. The firm that absorbs every AI efficiency as a discount needs an ever-larger volume of work just to hold revenue flat. Same team, same talent, more deals required every quarter to stand still.

## Why the obvious fixes fail

The two reflexive responses both collapse. Raise rates, and you invite the comparison shopping that hourly billing already encourages. Keep billing hours and quietly pocket the AI savings, and you are betting clients never notice, which is not a bet worth making.

The real answer has been sitting in plain view for over a decade. Ron Baker and others have argued since the 2000s that firms should price the value to the client, not the inputs. Nobody buys an hour. They buy an outcome. The advice was sound and mostly ignored, because hourly billing was comfortable and clients kept paying for it.

AI is what finally removes the comfort. Before, Erik said, it was easy not to make the shift, because clients weren't asking for it. "They might say they need an outcome, but they're used to looking at the value of that outcome in terms of hours and rates, and it's a tough conversation to shift their thinking to value. But if we have no choice but to do it, then that makes it a little bit easier. AI is like this forcing function."

The pressure is not a reason to retreat. It is what finally makes a change firms should have made years ago achievable.

## The reframe: price the deliverable, not the hours

The move is to make fixed-fee the default and sell the deliverable instead of the time. Hours still inform the estimate. They stop being the product. This is what the pricing world calls value-based pricing, and at its furthest edge, performance-based pricing tied to the result the client actually gets.

"With fixed fee we don't need to be concerned quite so much with hours and times," Erik said. "It's truly almost value based. The underlying hours are a foundation, but they're not the only factor that go into it." Once the deliverable is the contract, an efficiency gain stops being a discount handed to the client by default. It becomes shared upside.

For Andrew, who sits close to the sales conversation, the value shows up first as speed. "How fast can we give them value," he said. AI helps a team reach a functional prototype early, something the client can see and react to. "They start feeling like I'm getting something, and I'm getting something quickly, and they can see that value continue to grow over time." That early proof reframes the whole conversation away from headcount and hours and toward what the client is actually buying. The teams that pull this off well tend to be the ones that already ship fast, which is its own argument for [an architecture built for speed across platforms](/our-services/application-development/).

A Claude-enabled workflow took [Divine's vibe-coded Flutter app to a production-ready release in 90 days](/success-stories/divine/), with 500+ issues addressed along the way. Rabble, Divine's founder and CEO, put the result plainly. "They integrated Claude into the workflow in a way that actually improved velocity and quality at the same time."

## The mechanism: share the savings, take the math out of the room

Pricing the deliverable raises an immediate, fair question. If AI made the work cheaper to deliver, why should the client not see all of that saving?

The answer is that they see most of it, not all of it. Erik framed it as the conversation he'd have with a client. "We're going to be able to do this 30% faster because of AI. We're going to pass some of that savings on to you." The majority of the gain goes to the client as a visibly better price. The rest covers real costs like token usage, hedges against the uncertainty of how much AI actually speeds a given job, and leaves room for margin.

Erik is candid about the limit. "If we were to just pass all of the AI efficiency over to the client as a discount every single time, then it defeats the purpose." Share enough to win against firms still billing the old way. Keep enough that getting better pays you something.

Andrew's emphasis is on how that reaches the client, which is not as arithmetic. "We're not going to give them that math," he said. "We're going to frame it as, we think we're going to get efficiency via these great tools the team's built, so we'll give you a discount on that. Taking the math out of the equation and giving them just what the value is." This is not about hiding the calculation. The client is buying an outcome at a price, and handing them a spreadsheet of hours and efficiency percentages just invites a negotiation about inputs, which is the exact conversation pricing the deliverable is meant to end. What the client wants to know is what they get and what it costs.

## The catch nobody puts on the brochure

Better pricing does not, on its own, make the business more profitable. It can quietly make it worse.

When AI shortens a project, the people on it finish early. That time does not disappear from the payroll. "That creates a gap," Andrew said. "What are those folks doing for that additional time?" If the answer is nothing, the firm has not captured an efficiency. It has bought a more expensive bench.

The pricing model surfaces the gap rather than hiding it. When a deal is priced with AI efficiency built in, the tooling flags the freed capacity and signals the business development team. "We need to sell to fill that capacity," Andrew said, "not creating that bleed on the organization by creating unused bench capacity." Erik put the cost plainly. "If they're just sitting idle, then there's an opportunity cost there. Resourcing is just a constant puzzle that never ends."

AI efficiency is, in large part, a sales problem wearing a pricing costume. You can get the pricing exactly right and still end up with idle people on the bench. When asked directly whether that was true, Andrew did not dodge. "I think that's true."

So what fills the gap? Not a pricing tweak. New work. Fixed-fee plus visible AI efficiency is a stronger pitch than another rate card. But it only pays off if the pipeline grows to absorb the capacity that efficiency frees up. That's a new demand the hourly model never placed on anyone. Pricing exposes the problem. Sales has to solve it.

That is the bind. Andrew resists the idea that it dooms anyone. "There's probably some truth to that," he said of the notion that getting faster makes a services firm poorer, "but the way we manage that problem is going to dictate if we become poor or not." AI does not only shrink a project. It also lets a team carry more projects at once. "Instead of one project you can have agents working on multiple projects." Freed capacity is a threat only when it sits idle. Put to work, it is leverage.

## The harder change is cultural

The mechanics are the easy part. The shift underneath them is in how a firm relates to its clients.

Hourly billing sets the tone for that relationship, and the tone is transactional. As Erik described it, when the thing you deliver is measured in time sheets, you get treated like a vendor, whatever language sits on top. He caught himself even saying it, conscious that his own teams deliver far more than billable time. The point stands anyway. When the deliverable is an outcome, the relationship changes. "We need to be less reactive to a client and just more consultative," he said. He called it the muscle we most need to exercise.

Andrew sees the same muscle on the delivery side. The new discipline is "showing that value consistently week over week over week," whether that is feature demos or updates on live product. "That's a muscle that's got to be developed, and the team's got to get really comfortable with that level of outputs that are visible." Visible, repeated proof of value is what makes a fixed price feel earned rather than asserted.

## What we have not figured out yet

The honest frontier is measurement. The entire model leans on knowing how much AI actually speeds a given kind of work, and nobody in the industry has that number nailed down yet. The efficiency is not uniform. It lands differently across design, engineering, QA, and program work. As Andrew put it, "It's lumpy. It's not very flat yet." Until that flattens into something repeatable, the percentages in any quote carry real uncertainty, which is exactly why the model shares savings conservatively and hedges the rest. And because the fee is fixed, the client is never exposed to the estimate. If a job takes longer than expected, that is VGV's risk, not the client's bill. The numbers get sharper as real delivery data accumulates.

Getting there is its own project. It means tracking estimated against actual delivery on every engagement, standardizing tooling so teams are not each measuring a different thing, and sharing what works across the organization so the efficiency is consistent enough to price against. None of that is solved. All of it is being built.

For any services leader still billing purely by the hour, the takeaway is not a finished blueprint. It is a place to start. Andrew's version is blunt. Look at what your teams produce and ask whether the value is something you can actually see. "If the answer to that is I don't know or no," he said, "then we need to think about how we can get them value with a known cost and a known team." A team that delivers faster against a guaranteed outcome, he argues, beats one you measure in hours.

The deeper risk is standing still. "The biggest risk right now is complacency," Erik said. The specifics of AI's impact are not fully clear to anyone, including us. What is clear is that continuing exactly as before is the one option that does not survive contact with the next few years. As Erik put it, whatever you do on Monday, it should probably be something other than what you are doing now.

*Working through how AI changes the economics of how you build and ship? That's [a conversation we're having every day](/contact-us/), and it is the work our [AI and strategy workshops](/our-services/ai-strategy-workshops/) are built for.*