Most statements of work are written to be signed, not read. Twelve pages, a liability section copied from the last one, an hourly estimate per feature that nobody believes, and a signature block. The buyer skims to the number, the seller hopes they do, and the document goes into a folder until something goes wrong.
Ours is two pages, and it arrives three days after the first call, before any money has changed hands. That is not a courtesy. It is the part of the sale that does the selling, because it is the first piece of work you see from us, and it is work you can check.
What goes in
Page one is four lists. In: the first increment, written as things a user can do, not as components. Out: everything we discussed that is not in the first increment, by name. Cost: one fixed number for the increment, with the day rate it was built from, so you can do the arithmetic yourself. Acceptance: how we both know an item is done, in your environment, before the invoice.
Page two is how it runs. The fortnightly review and what it reports: shipped, slipped, cost against plan. Who on your side signs the increment. What happens when scope changes, which it will. And what stops the work: the two or three conditions under which either of us can call it without a conversation about fault.
The out list is the page a CTO reads twice
Anyone can write the in list; it is the brief, reformatted. The out list is where the thinking shows. It says we understood the whole product, chose a first slice on purpose, and are not going to let the increment quietly grow until the budget is gone. When a head of engineering reads single sign-on: out, increment two, they learn two things at once: we heard it, and we are not pretending it is free.
A scope that lists only what is in has made no decisions. It has copied the brief and added a price.
What we leave out on purpose
- Hour estimates per feature. They are precise, wrong, and used later as a weapon. The increment has one number.
- A change-request process with forms. Changes go into the next increment, and the fortnightly review is where they are agreed.
- Velocity promises. We commit to a scope and a date, not to points per sprint.
- Liability boilerplate that pretends a two-page scope is a forty-page contract. The contract is a separate document, and it is short too.
- Anything we have not shipped before. If a line would be our first time, it says so, or it is not there.
Why before the money
Because it is the cheapest test either side will ever run. Three days of our thinking, free, and you learn whether we understood the problem, whether we can write, and whether we will say no to anything. If the two pages are vague, the next six weeks will be too. If they are sharp, you are holding a document you could hand to any other firm and get a comparable quote, which is the point: a scope you can take elsewhere is a scope you can trust.
How to read one, whoever wrote it
- Can you find the first increment without asking? If the answer is everything, in phases, it is a brief, not a scope.
- Does it name what is out? If nothing is out, nothing has been decided.
- Could you check an acceptance line yourself, in your own environment, with nobody from the vendor in the room? If not, done means whatever they say it means.
Ours passes those three or it goes back for another day. That is the whole pitch.