What is a Master Services Agreement?

Most services relationships are governed by two documents doing two different jobs, and most contractual trouble begins when one of them quietly starts doing the other’s. The Master Services Agreement holds the relationship. The Statement of Work holds the work. Keeping them apart sounds like legal housekeeping. It is actually one of the more consequential operating disciplines a services firm has.

The container and the work

A Master Services Agreement is the container contract: the durable frame within which every future piece of work will sit. It records the terms both parties want settled once, for everything: liability caps and indemnities, intellectual property ownership, data protection and confidentiality, insurance, payment terms, termination rights, dispute resolution. It deliberately contains no scope. It describes no project, prices no work, promises no dates.

The work lives in Statements of Work signed under the MSA. Each SOW describes one engagement: scope, deliverables, rates, milestones, acceptance criteria, timelines. The design intent is division of labour. Negotiate the legal relationship once, slowly and carefully, with lawyers on both sides. Then let each new engagement start on a short document that only has to describe the work, because everything else is inherited from the master.

What belongs where

The sorting rule is simple to state. Anything that should hold true regardless of the specific project belongs in the MSA: how much liability the supplier holds, who owns what gets created, how each side treats the other’s data and confidential information, how the relationship ends, how disagreements get resolved. Anything that genuinely varies from engagement to engagement belongs in the SOW: what is being built or run, by whom, at what rates, delivered when, accepted how.

The test for a clause is whether it should survive this particular piece of work. An IP ownership position should not depend on which project team drafted the paperwork. A milestone date obviously should. Firms get into trouble not because the rule is hard to understand but because the two documents are drafted by different people under different pressures: the MSA by lawyers with time to think, the SOW by commercial and delivery people trying to close and start work.

The drift: when SOWs quietly amend the master

That difference in drafting pressure is where the drift begins. A client procurement team pastes its own acceptance language into a SOW, and the acceptance regime now differs from the master. An account lead concedes bespoke liability wording on one urgent deal. A delivery team accepts an IP carve-out for a single deliverable. Payment terms slip on one SOW because the quarter needed the signature. None of these are labelled amendments. Formally, most MSAs even say a SOW cannot vary the master without explicit reference. In practice, each concession sits in a signed document, and signed documents have a way of becoming the operative truth.

The result is a firm whose stated risk position lives in the MSA and whose actual risk position lives in the accumulated exceptions across dozens of SOWs, uncompared and uncounted. Order-of-precedence clauses help in a courtroom. They do not help the executive who believes the firm’s liability is capped in a way that, on three specific engagements, it no longer is. An MSA is the firm’s risk position written down once; every SOW that quietly rewrites it is an unpriced decision to hold more risk.

One concrete example

Clearly illustrative, with no customer implied. A consultancy signs an MSA with a client that caps liability at the value of fees paid under the relevant SOW. Over three years the relationship grows to a dozen live SOWs, drafted by different account teams. On one of them, signed late in a quarter, the client’s template acceptance schedule was pasted in wholesale, and buried within it is uncapped liability for data incidents. Nobody priced that. Nobody escalated it, because it arrived as an annex to a routine work order, not as an amendment to the master. The firm’s leadership would say, honestly and wrongly, that its exposure on the account is capped. The gap between what they believe and what is signed is not a legal technicality. It is an unrecorded decision, made under deadline pressure by someone who may not have known they were making it.

A contract is a set of recorded decisions

The decision-intelligence reading of the MSA starts from what the document really is: a set of decisions about risk, made deliberately and recorded in one governed place. How much liability to hold, what to indemnify, who owns the work product. Seen that way, the drift problem comes into focus. Every SOW deviation from the master is also a decision, with options (hold the standard, concede with a price attached, escalate) and a cost, and the failure mode is that these decisions are made silently, by whoever happens to be drafting, without being raised, priced or recorded. The discipline that fixes it is the same one that governs scope through change orders: deviations from the agreed position are surfaced as decisions, taken by someone with authority to take them, and written down where the next person can find them. That is the decision-intelligence view of contracting. The MSA is not paperwork that precedes the work. It is the firm’s judgement about risk, made durable, and it only stays true if the organisation notices when it is being quietly rewritten.