When the Contract Moves Faster Than the Cost Base

Outcome-based pricing makes revenue variable while a permanent delivery bench keeps cost fixed, and the gap between those two facts is where margin disappears. This is the structural problem underneath the entire pricing transition, and it is more consequential for a mid-sized firm than any individual contract term.

The arithmetic

Under time and materials, revenue and cost move together. A person on a project generates revenue; a person off one stops, and utilisation reporting makes the gap visible within days.

Under outcome pricing that link breaks, and it breaks in both directions. Revenue attaches to a result delivered on the client's timetable. Cost attaches to people employed on the provider's payroll. A contract that pays on measured improvement in month nine still requires a fully staffed team in months one through eight.

Three further characteristics of AI-era deals sharpen the problem. Contracts are reported to be smaller and shorter than those they replace, which compresses the period over which fixed capacity can be amortised. Payment is often back-loaded, as in the multiyear cloud management agreement HCLTech signed with E.ON in June 2025, where no compensation flows in the first 12 months. And the skills required change faster than a permanent workforce can be retrained, so a team optimised for last year's delivery model is not automatically the team that hits this year's outcome.

The scale of the squeeze is visible in the 2026 numbers. The Nifty IT index fell 20% across the year, with 10 constituents shedding a combined $73 billion in market value, while Persistent Systems reported clients asking for the same scope at 25% to 30% less. A firm absorbing a 25% price cut on a cost base that has not moved has a structural problem, not a margin one.

The options, honestly costed

No arrangement removes the mismatch. Four manage it, each with a real price.

The fourth option deserves attention because it is the instinctive response to margin pressure and the most dangerous. A firm that strips its bench to protect this quarter's margin recreates the mobilisation delay that used to be the industry's differentiator, except that clients now expect speed and will not pay a premium for it.

Core-and-flex, in practice

Most firms that have made this work have landed on some version of a core-and-flex structure, and the design questions it raises are unusually specific.

What belongs in the core? Client relationship depth, domain knowledge, delivery leadership, architecture, and the capability the firm competes on. These are the things that take years to build and cannot be acquired at short notice.

What belongs in the flex? Specialist skills needed in bursts, capacity for volume peaks, and capability the firm needs occasionally but cannot keep utilised. A machine-learning engineer needed for eleven weeks a year is a flex requirement regardless of how strategically important those eleven weeks are.

What is the ratio? There is no industry answer, but the question can be derived. Take the portfolio, identify revenue that is genuinely committed for more than twelve months, and treat that as the maximum defensible permanent capacity. Everything above it is a bet on renewal.

Most mid-sized firms discover their permanent headcount implies confidence in revenue they do not actually have contracted.

What does the flex cost? Contracted access to capability carries a margin share that looks expensive against an employed rate card. The comparison is misleading, because the employed rate card excludes the cost of carrying that person through non-billable periods. The correct comparison is flex cost against fully-loaded annual cost divided by realistic utilisation, and on that basis the gap narrows sharply.

This is the practical case for treating capability as something a firm accesses rather than owns — the argument behind managed capability models generally, and one worth evaluating on the arithmetic rather than the positioning.

What to change first

Three moves, taken in order, do most of the work.

Measure cost-to-serve an outcome, not a person. Most firms know their cost per person-month precisely and their cost per resolved case not at all, and outcome pricing requires the second number. Until it exists, every outcome contract is priced on hope rather than on arithmetic.

Separate committed revenue from expected revenue in capacity planning. Permanent headcount should be sized against contracted revenue, not forecast revenue. Firms routinely plan against the latter and then describe the consequence as a bench problem when renewals slip.

Build the flex relationship before it is needed. A partner arrangement negotiated under pressure, mid-contract, costs more and delivers worse than one established in advance. What is being purchased in a standing arrangement is lead time on access to scarce capability, which is precisely the thing that cannot be bought quickly.

The measure worth watching

Utilisation was the right metric for a time-and-materials business. It answers a question that matters less each quarter: what proportion of the people we employ are billing.

The useful replacement is a cost-to-outcome ratio — what it costs to deliver one unit of the thing clients now buy, tracked over time. A firm improving that number is genuinely getting better at the new model. A firm with excellent utilisation and a flat cost-to-outcome ratio is running the old business efficiently while the market moves.

Persistent grew revenue 16% and Coforge 33% in the second quarter of 2026 while TCS, Infosys, Wipro and HCLTech managed 1–3%. Size was not the differentiator between those two groups; the ability to reconfigure a delivery model at speed was, and that ability is largely a function of how much of the cost base can actually move.