Statement of work vs contract, and how they connect.
The contract governs the relationship. The statement of work governs the project. Four short clauses decide whether the pair actually holds together when a client asks for something you never agreed to build.
A contract sets the terms between two businesses, meaning payment, ownership of the work, confidentiality, liability and how either side gets out. A statement of work sets one project inside those terms, meaning the deliverables, the schedule, the fee and what counts as finished. Almost every agency engagement needs both jobs done, though not necessarily in two separate files, and the part that decides whether they work is a handful of sentences most agencies never write. Skip those sentences and you end up with a contract that carefully protects a project nobody actually defined.
That gap is where the expensive arguments start. The terms document says the client owns the deliverables on final payment, the scope document lists eleven pages, and in month three somebody asks for a twelfth page while pointing at whichever line suits them. Sorting out which document governs that request after the fact is a much worse conversation than writing the four clauses below before the project starts.
What's the difference between a statement of work and a contract?
The contract governs the relationship and the statement of work governs the project. Your contract covers the things that stay true across every engagement you ever run with that client, so payment terms, late fees, who owns what and when, confidentiality, indemnity, limitation of liability, and the notice period either side needs to walk away. Your statement of work covers the things that change with every project, so the list of deliverables, the dates, the dollar amount, the number of revision rounds, and the conditions under which a deliverable is accepted rather than merely delivered.
A useful way to hold the distinction is that the contract answers what happens if this goes wrong, while the statement of work answers what exactly we agreed to build. A carefully drafted contract with no statement of work establishes who is liable without establishing what they were liable for, and a detailed statement of work with no terms behind it describes eleven pages of work with no stated remedy when the invoice goes unpaid for ninety days.
None of this requires two PDFs, and the $24,000 question below is where the one-document shape gets decided. If you're still untangling the neighboring vocabulary, the difference between the scope of work and the statement of work sitting around it is worth ten minutes on its own.
Is a statement of work legally binding?
A statement of work signed by both parties, with a fee and a schedule in it and nothing else governing the work, is generally treated as a legally binding agreement in its own right in ordinary US business practice. One issued under an existing master agreement is normally read together with that agreement instead, so the pair binds rather than the single file.
Which is to say it depends on how the document was issued, and that's why the wiring clauses below matter more than the label at the top of the page. We build software for agencies rather than practicing law, so treat all four clauses here as the shape of a document rather than advice about your situation, and have a lawyer read your standard template once.
Do you need both on a $24,000 project?
You need both jobs done on a $24,000 project, and for most agencies at that size the cleanest shape is one signed statement of work that carries its own terms section. Splitting a six-week engagement into a master agreement plus a statement of work adds a signature round and a week of back and forth for a client who was ready to start on Monday. The terms still have to be in there somewhere, so the statement of work grows a section covering payment, ownership, confidentiality and termination, and the client signs one thing.
The shape changes once a client comes back. By the third project with the same company you're negotiating the same liability language for the third time, which is where a master services agreement plus a short statement of work per project starts paying for itself, and we've written up how those two documents divide the work in more detail.
The failure mode we see most often sits between those two shapes. An agency sends a polished proposal with the scope and the price in it, the client says yes over email, and a separate terms PDF gets signed a week later with no mention of the proposal anywhere in it. Two documents now exist and neither one says the other applies, which means the scope you spent four hours writing is arguably just sales material. Sorting out where a proposal ends and a statement of work begins is the fix, and the clause below is the mechanism.
Which clause actually connects the two documents?
Incorporation by reference is the clause that connects them, and it belongs at the top of every statement of work you send. It's one short paragraph that names the contract, names its date, and states that the terms of that contract apply to the work described below. Without it you have two signed documents in a folder with no stated relationship, and any argument about which one applies starts from zero.
The connection is worth making in both directions, so your contract should say somewhere that work will be performed under statements of work issued and signed from time to time. That sentence costs nothing and closes the reverse gap, where a client's finance team reads a master agreement, sees no dollar figure in it, and concludes that nothing has been committed to.
One detail catches agencies out here. A master agreement signed in January doesn't commit anybody to a project scoped in September, so each statement of work needs its own countersignature rather than riding along on the original one. The signature on a specific statement of work is what commits both sides to that specific fee and that specific list of deliverables.
What happens when the contract and the statement of work disagree?
An order of precedence clause decides, and if you never wrote one then the disagreement gets settled by whoever argues longer. The conflict is more common than it sounds, because the two documents are written months apart by different people with different concerns. Your contract template says two rounds of revisions on any deliverable, the statement of work you wrote last Tuesday promises three, and the client has now used all three and is asking about a fourth while quoting your own paperwork back at you.
That split is the right default for agency work, and it's worth understanding why rather than copying it blindly. The contract carries the legal terms that were negotiated carefully once and shouldn't quietly change because somebody was writing a scope at eleven at night. The statement of work carries the commercial specifics that were agreed most recently and most deliberately for this project, so it should win on exactly those specifics. Any clause that hands total precedence to the contract effectively lets an old document overrule a fee you negotiated last week.
Prevention is the other half of the fix, since most precedence conflicts exist only because the same subject got written twice. Keep revision limits, deliverable lists, payment schedules and acceptance rules out of the contract template entirely and let the statement of work own them, because a clause that resolves a conflict is a worse outcome than never having one.
Where does a change order live, the contract or the statement of work?
The change control clause lives in the contract, and the change order it produces amends the statement of work. That division confuses people, so it's worth stating plainly. The rule about how changes get agreed is a relationship term that shouldn't be renegotiated per project, while the actual change, meaning the extra page and the extra $1,800 and the new date, is a project fact that belongs with the other project facts.
That last sentence is the one that earns its place. Without it, your paperwork says changes need approval while your actual practice is that somebody starts on Tuesday and raises the money question on Friday once it's half built. Saying out loud that work begins after approval turns an awkward conversation into a procedural one, and it's far easier to decline starting something than to invoice for something already delivered.
An approved change carries the same weight as the original signature, which is exactly why the approval needs to look like a signature rather than a thumbs up in a channel. A short signed change order naming the deliverable, the new fee and the new date is enough, and our change request form template lays out the fields that matter. Agencies that skip this step tend to underestimate what it adds up to, and the arithmetic behind what scope creep costs over a year is genuinely uncomfortable reading.
How do you word acceptance so "done" isn't a negotiation?
Acceptance needs a review window, a written rejection requirement and a default for silence, and all three belong in the statement of work rather than the contract. Delivery and acceptance are different events, and the gap between them is where projects go to die. You send the eleven pages, the client says thanks, and nothing else happens for five weeks while your final invoice sits unbilled because the project was never formally finished.
The deemed acceptance default is the load bearing part, because client silence is far more common than client rejection. Five business days reads as fair rather than aggressive, which matters because a clause a client refuses to sign protects nobody. The final sentence quietly does a second job by routing new requests into the change process instead of letting them arrive disguised as bug reports, which is the single most common way an extra $4,000 of work gets absorbed for free.
Acceptance criteria only work if the deliverables are countable and testable, which is a writing problem rather than a legal one. A line saying a homepage is accepted when it's complete gives you nothing, while one saying it's accepted when it's live on the production domain and matches the approved design at three screen widths gives both sides the same answer, and how to write a scope of work clients can't misread covers that mechanic properly.
The clause that lets an old document overrule last week's fee
Any clause handing total precedence to the contract does exactly that, because a revision limit typed into a template in January then quietly outranks the fee and the deliverable list you agreed for this project last week. Open your most recent statement of work and read it against the four clauses above. It should name the contract it's issued under and the date of it, say which document controls on scope and fees, state when work on a change may begin, and give acceptance a deadline and a default.
Most agencies find two or three of the four missing, and fixing them is an afternoon of work on a template you'll reuse for years rather than an ongoing cost. Write the four clauses yourself in plain language, then pay a lawyer for one hour to read the finished template rather than to draft it from scratch.
A contract that protects a project nobody defined is an expensive kind of comfort.
The clauses themselves are only half the problem, because a change control paragraph is worth exactly as much as the record it produces. Docket exists for that second half, turning each change into a request the client approves against a stated price and date, with the approval, the timestamp and the exact wording kept together and exportable as a signed PDF. A record of approvals doesn't decide which document controls or what a liability cap covers, since that's the job of the four clauses above, but it answers the question those clauses exist to answer, which is what both sides agreed to and on what date.
Frequently asked questions
Can you send a statement of work without a contract?
You can, and plenty of small agencies do, but the statement of work then has to carry every term you would otherwise have relied on. That means the payment schedule, the late fee, who owns the files before and after final payment, and how either side walks away. A single signed document works fine for a one-off engagement. Once a client becomes a repeat client, splitting the terms into a master agreement saves rewriting them every time.
Which document wins if the statement of work and the contract disagree?
Whichever one your order of precedence clause says wins, and if you never wrote that clause then the answer is an argument you have in month three. The usual arrangement for agency work is that the contract controls on legal terms while the statement of work controls on scope, deliverables, fees, schedule and acceptance. That split matches how the two documents actually get used over time.
Does every statement of work need its own signature?
Yes, and it is the cheapest protection available to an agency. A master agreement signed in January does not by itself commit anyone to a project scoped in September, so each statement of work needs its own countersignature against that specific fee and that specific list of deliverables. The same logic applies to every change afterward, which is why a change order needs an approval of its own rather than inheriting the signature on the original document.
Do small agencies actually need a master services agreement?
Not for a single one-off project, where a well written statement of work carrying its own terms is usually enough. The calculation changes once a client sends a second project, because you then negotiate the same liability cap and the same ownership language twice for no reason. Roughly three expected engagements with the same client is where the framework starts paying for itself.