When a client asks what a project will cost and we say "somewhere between 480 and 720 hours", the first reaction is usually mild frustration. They wanted a number. What they got looks like we are hedging.
We are not hedging. The range is the honest version of the answer, and the single number would have been the dishonest one.
A point estimate pretends the unknowns don't exist
Every estimate is built on assumptions: that the third-party API behaves the way its documentation claims, that the existing data is as clean as we were told, that "a simple admin screen" means the same thing to both of us. Some of those assumptions will be wrong. We just don't know which ones yet.
A single number quietly bakes in a guess about all of that and then hides the guess. A range keeps it visible. The bottom of the range is roughly what it costs if the assumptions hold; the top is roughly what it costs if the ones we are most worried about don't.
The width of the range is information
A tight range — 10 or 15 percent — means we have built something close to this before and the unknowns are small. A wide one — 40 percent or more — means there is a real question mark somewhere, and the range is us pointing at it.
That is the useful part. If the range is wide because of one integration nobody has scoped properly, we can spend a few days scoping it and the range gets narrower. You are buying down uncertainty with a small, early piece of work instead of discovering the cost of it halfway through the build.
How we actually use it
We estimate each area of the project separately — auth, the API, the admin tools, the reporting — and give each one its own range. Then we are explicit about which ranges are wide and why.
On most projects, two or three areas carry almost all of the uncertainty and the rest are well understood. We would rather show you that than average it all into one confident-looking figure that is wrong in both directions at once.
What we commit to
Once the wide areas are scoped down — usually in the first week or two — the range tightens and we will hold to the top of the tightened range as a not-to-exceed figure for that scope. If something outside the written scope comes up, it goes through the change process and gets its own small estimate before anyone builds it.
That is the trade: you don't get a single number on day one, but the number you do get a fortnight later is one we will stand behind.