How to read an estimate you did not write

First: the front matter, not the numbers

Open the assumptions and exclusions page. If there is not one, that is your finding, and it is a serious one. An estimate without a written scope statement is a number with no boundary, and it will be read as covering everything by whoever receives it.

What you are looking for is not whether the exclusions are reasonable, it is whether anything excluded is actually your problem. An estimate that excludes temporary works is fine if a separate package covers them, and dangerous if nobody has noticed.

• Which drawings, at which revision, on which date
• What is explicitly excluded, and who is carrying each excluded item
• What the estimator could not resolve from the drawings and assumed instead
• Whether the assumptions on the tricky parts match what you would have assumed
• Which rates are quoted and which came from a library

Second: the shape of the number

Before any line detail, take the total and divide it by something. Square metres of floor, linear metres of facade, number of beds, number of parking spaces. Then compare that against a job you have actually built.

This one check catches more than everything else combined, because almost every serious error makes the building come out the wrong size. Scale errors, missing sheets, a dropped package and a duplicated one all show up here, and none of them show up in a line-by-line read.

Then do the same by division. If concrete is thirty-one percent of the total on this job and it was nineteen percent on the last three like it, that is a question worth an hour of somebody time.

Third: sort by value and read down

Sort the whole workbook by line total, descending, and read from the top until the lines stop mattering. On a typical building estimate somewhere between forty and eighty lines carry eighty percent of the cost. Those are the ones where a fifteen percent error changes your bid position.

For each of those lines, three questions. Is the quantity plausible against the drawing? Is the unit rate one you would pay today? Is the waste factor sensible for that material and this geometry? You are not re-estimating, you are sanity-checking the things big enough to hurt.

Fourth: hunt for the specific failures

Now search rather than read. Sort or filter for the patterns that indicate a problem instead of reading for comprehension.

1. Zero quantities and zero rates. Usually a placeholder that never got filled in.
2. Quantities that are suspiciously round. Real buildings produce 487, not 500.
3. Lines with no unit, which usually means nobody could decide how to measure it.
4. Identical rates on lines that should differ, a sign a rate was copied down a column.
5. Provisional sums and allowances, added up together. If they exceed about eight percent of the total, the estimate is carrying more uncertainty than it appears to.

Last: the arithmetic

Cross-foot the divisions against the total, check that the markup applies to what it is supposed to apply to, and confirm nothing is double-counted between a division and a preliminaries line. This is genuinely worth doing, and it is genuinely the least likely thing to be wrong, which is why it goes at the end rather than the beginning.

The arithmetic is almost always right. The scope almost always has a hole in it.

The forty minute version

1. Five minutes: exclusions and assumptions.
2. Five minutes: total per unit against a benchmark job, then the same by division.
3. Twenty minutes: the top forty lines by value.
4. Five minutes: search for zeros, round numbers, missing units, repeated rates.
5. Five minutes: cross-foot, and check where the markup lands.

That will not make you certain. It will tell you whether to sign it, and it will tell you which three things to ask about, which is what you actually needed.