Who Owns the Number?

Once a services contract pays against a business metric, the provider becomes exposed to everything that moves that metric, including decisions made entirely inside the client organisation. Defining which part of the number the provider owns is not a legal formality; it is what separates a contract that can be delivered from one that cannot.

The exposure nobody prices

Consider a provider paid on a 25% reduction in customer service handling time. It redesigns the workflow, deploys automation and retrains the support team's tooling. Handling time does not move, because the client's product team shipped a release generating a new class of complex queries and the client's own agents were never given time to adopt the new process. The provider did the work; the number did not move; under a poorly drafted contract the provider is not paid.

This is the most common way outcome contracts sour, and it damages the relationship rather than merely the commercials, because both parties believe they behaved correctly.

The exposure matters more each quarter because the contracts carrying it are multiplying. Cognizant signs roughly 45% of its BPO contracts on outcome-based models, and around 80% of TCS contracts in its finance, HR and business-services segment now carry outcome performance measures, double the proportion of late 2023. Reporting on Indian IT in August 2026 identified provider capacity to absorb financial risk as the main brake on adoption, and much of that capacity question is not balance sheet at all. It is whether the provider drew the accountability line precisely enough to know what it accepted.

The line that holds

There is a defensible boundary, and it is worth stating precisely because commercial pressure erodes it.

A provider can reasonably commit to four things:

• Capability: the skills and capacity required, available when required.
• Availability: service levels, response times, coverage windows.
• Throughput: volume processed, cases handled, features deployed.
• Quality: defect rates, rework, accuracy, compliance with defined standards.

A provider cannot reasonably commit to four others:

• Adoption: whether the client's people use what was built.
• Scope discipline: whether requirements stay stable through the term.
• Data quality: the state of inputs the provider does not generate.
• Business results: revenue, market share and customer satisfaction, which depend on pricing, product, competition and a dozen factors outside the engagement.

The second list is not a list of excuses. Each item describes a lever held by the client. A provider who accepts accountability for them has not been bold; it has been imprecise, and imprecision in a contract is a liability that arrives later.

Why Tech Mahindra's metric set is worth studying

The healthcare engagement Tech Mahindra won, priced against a 40% reduction in tickets, a 20% reduction in resolution time and a 30–35% reduction in technical debt, is instructive for what it does not contain.

None of those three measures is a business result. Each is operational, each sits substantially within the provider's influence, and each can be measured from systems both parties can read. The contract carries genuine risk — missing any of them has consequences — but the risk is bounded by the provider's own performance rather than by the client's decisions.

Compare that to a contract paying on patient satisfaction or cost per episode of care. Those are the outcomes the client ultimately wants, and they are also shaped by clinical staffing, payer behaviour and regulation. A provider carrying that risk is being paid to influence a system it does not control.

The discipline is to price against the operational measures that drive the business result, not against the business result itself.

Writing it down

Four mechanisms make the boundary enforceable rather than aspirational.

Dependency clauses. Named conditions that must hold for the measurement to apply: client staffing at agreed levels, data quality above a defined threshold, adoption above a stated percentage, scope frozen or formally repriced. Each should specify the consequence if it fails, whether that is suspension of the measure, an adjusted target, or a fallback to a time-based fee.

Change control with teeth. Outcome contracts fail through accumulated small scope additions more often than through single large ones, so the mechanism has to be light enough that people use it and firm enough that it is not bypassed.

A monthly joint governance forum. Both parties' operational leads reviewing the metric and its dependencies together, to surface drift while it is still correctable rather than at the annual true-up.

A named owner on each side. Every measure should have one accountable person at the provider and one at the client. Metrics without named owners become nobody's problem until they become everybody's dispute.

The conversation to have out loud

The most useful thing a provider can do in a negotiation is state the boundary explicitly rather than leaving it to drafting. Something close to:

We will commit to the capability being available, to the throughput, and to the quality standard. We will work alongside you on adoption and report on it. We cannot be paid or not paid on whether your teams adopt it, because we do not control that lever, and a contract that pretends otherwise will damage the relationship in month nine.

Clients respond better to that than providers expect, and a buyer who has run a failed outcome contract before recognises the honesty immediately. The alternative, accepting unbounded outcome accountability to win the deal, purchases a signature at the cost of a dispute that arrives later, costs more, and usually ends the account.

The governance point most firms miss

There is a version of this that goes wrong internally rather than contractually. When a firm signs its first metric-linked contract, accountability inside the delivery organisation often stays where it always was, with a project manager measured on schedule and effort. The commercial model changed; the internal accountability model did not, and the team optimises for the old measures while the contract pays on new ones.

Someone in delivery has to own the contracted number, with the authority to change how the work is done in order to move it. Without that, the firm has taken commercial risk it has no internal mechanism to manage.