Outcome-Based Pricing Is Replacing IT Billable Hours

Ask any services CEO whether they sell time or outcomes, and every one of them will say outcomes. Then look at the invoice. It is still a headcount, multiplied by a rate, multiplied by a month — the same arithmetic we have used for thirty years. That gap between what we claim to sell and what we actually bill is where this industry is being repriced right now, and FY26 finally gave us numbers instead of opinions.

I have spent twenty-five years on the delivery side of that invoice, mostly in Gujarat, mostly on enterprise programmes that were quoted in person-months. I have also priced outcome deals badly enough to lose money on two of them. This is what the last quarter changed, what I got wrong, and what I would do differently.

The FY26 numbers that ended the headcount era

Revenue per employee is the metric that exposes this shift, and it moved. TCS, Infosys and HCLTech all posted a 3–4% rise in revenue per employee in FY26, while tier-1 Indian IT shed roughly 7,000 people net.

Read that again. Revenue per head went up while heads went down. In the linear model those two lines move together, because the only way to bill more was to staff more. They have separated.

TCS managed +3.4% revenue per employee while cutting 2% of its workforce, and its annualised AI revenue reached $1.8 billion, up 17.3% sequentially. HCLTech did better on the ratio, at +4.1%, and did it while adding people — which matters, because it shows the gain is not simply a subtraction trick.

The workforce shape tells the same story from another angle. TCS hired 40,000 freshers while releasing 12,000 mid-to-senior staff. Cognizant booked $230–320 million in severance under Project Leap and simultaneously committed to 20,000 freshers. Cheap juniors at the bottom, expensive specialists at the top, and the billable middle — the layer that existed to be staffed onto time-and-materials contracts — getting squeezed out.

Two ideas you have to accept before repricing anything

Nobody was ever buying your hours

The sharpest framing I have read on this came from Mark Wilson in an essay that surfaced on Hacker News on 25 July 2026: AI isn't killing consulting, it's killing time as a proxy for value. Clients never wanted person-months. They wanted a working migration, a passing audit, a reconciliation that stopped breaking. Hours were a proxy everyone tolerated because it was auditable and roughly correlated with effort.

AI broke the correlation, not the demand. When a task that consumed forty hours now consumes six, the proxy stops approximating anything. Billing the client forty is fraud. Billing six is a 85% revenue cut for identical value delivered. Neither is a business.

In 2023 we bid a data-platform migration for a mid-market insurer at a fixed price built on 2,100 person-hours. We delivered it in about 1,300 because two of our engineers had started using code assistants aggressively and the ingestion layer came together far faster than the estimate assumed. The client was delighted with the timeline and then asked, reasonably, why the invoice had not moved. I had no good answer, because I had priced the input while promising the output. We honoured the number and I spent the next quarter rewriting how we quote.

An outcome contract has four load-bearing parts

Every outcome deal I have seen fail was missing at least one of these. A stable base that covers your fixed delivery cost regardless of result. A clear target defined in the client's own reporting, not yours. Quality guardrails so the target cannot be hit in a way that damages something else. And shared upside that is generous enough to be worth the risk you just absorbed.

Drop the base and you have a lottery ticket. Drop the guardrails and you get a call-deflection target hit by making the support number harder to find. I have watched that exact thing happen.

Moving an existing contract without blowing up the relationship

You cannot reprice a whole book at once, and you should not try. Here is the sequence that has worked for us.

1. Pick the contract where the outcome is already instrumented. If the client cannot show you last quarter's number for the thing you plan to be paid on, you are negotiating in the dark. Start where a dashboard already exists.
2. Agree the baseline in writing before you touch anything. Thirty days of measurement, signed off by both sides. This is the step everyone skips and the step every dispute traces back to.
3. Keep 60–70% as a fixed base in year one. Put only the remainder at risk. You are buying evidence, not maximising upside.
4. Write the attribution rule down. Name what happens if the client's own marketing campaign, or a third vendor, plausibly caused part of the movement. Decide it while everyone is friendly.
5. Set a review date and a way out. Six months, either party can revert to the old structure. Nobody signs an experiment they cannot exit.

The buy side is already ahead of most vendors on this, which surprised me. A 2024 Deloitte study found 67% of consulting buyers now prefer fixed-fee arrangements over time-and-materials, up from 41% three years earlier. McKinsey reports that more than 30% of its global fees are already tied directly to client outcomes. Forrester put AI and outcome-based pricing at the centre of its State of Technology Services 2026. The demand arrived roughly three years before most delivery organisations were ready to price against it.

What success looks like, and where these deals actually die

Here is the part the pricing consultants underplay. McKinsey's own research finds that 70% of companies struggle to measure outcomes accurately in outcome-based contracts, and a Deloitte study found 45% of firms adopting outcome-based pricing hit margin pressure in the early period. That is not a reason to avoid the model. It is a reason to expect a bad first year and budget for it.

Attribution is where they die. Not pricing theory, not negotiation — attribution. When a number moves and three parties contributed, the contract either says who gets paid or it becomes a quarterly argument.

If you cannot name, in one sentence, who is at fault when the metric does not move, you have not finished writing the contract. You have written a disagreement with a signature block.

Track three things internally from day one. Gross margin per outcome deal against your equivalent time-and-materials work, measured quarterly rather than at close. The share of revenue that is not indexed to headcount — that single ratio is the honest scoreboard for this whole transition. And dispute frequency, because a model that pays well and generates an argument every quarter is not sustainable.

One caution on the disruption narrative. Microsoft has committed $2.5 billion, Amazon $1 billion, and OpenAI similar money to fund embedded implementation consultants, and it is tempting to read that as systems integrators being cut out. The evidence does not support that yet at the top end. Disintermediation is showing up clearly at the SMB and agency tier; there is still no clean, quantified case of a large enterprise firing its integrator to build a core platform in-house. If you serve the Fortune 500, you have time. If you serve the mid-market, you have considerably less.

What I would tell my 2023 self: the pricing model was never the hard part. Instrumentation was. We could not have priced an outcome deal well back then because we had no reliable measurement of what our own delivery actually produced, and you cannot sell a result you cannot count.

So the next conversation worth having with your team is not about pricing. Pull up your three largest accounts and ask a narrower question: for each one, could we state today, from the client's own systems, what changed because of us last quarter? If the answer is no on all three, that is your real project for the next two quarters — and it is worth more than any repricing exercise you could run in the meantime. I would start there.