Start with a sample plan
Choose the sample before opening the takeoff so the review does not drift toward the easiest rows. A practical plan samples every scope category, the largest quantity, and one row the assistant flagged as uncertain. For a small takeoff, check all rows; for a large one, use a fixed percentage and record the sample size.
The goal is evidence, not coverage theater. Record the sample in the review note with the row identifiers and the reason each row was chosen. The cost estimator service page describes the review support this step belongs to, and the throughput guide explains why a consistent sample makes later capacity comparisons meaningful.
Add the AACE recommended practices reference to the note when the review follows a formal estimating procedure. That reference gives the sample a shared basis without turning the review into a compliance exercise. Set the sample size before the first row is opened and write it in the review note.
A sample that grows after a bad row is found is no longer a sample, it is a search. For a package with more than fifty rows, ten percent plus every flagged row is a practical starting point. Adjust the size only when the correction history shows the package is stable.
Run three checks
The first check is the unit check: confirm each sampled row uses the unit stated in the brief. The second is the arithmetic check: recompute quantity times unit and compare it to the workbook formula. The third is the scope check: confirm the row belongs to the bid package and is not duplicated in another category.
These three checks catch most errors without re-measuring every line. Keep the checks independent so one mistake does not hide another. If the unit is wrong, fix the unit before testing the arithmetic.
If the scope is wrong, remove the row before checking the total. Reviewing a row that should not exist wastes the other checks and inflates confidence. Use a second screen or a printed sample sheet so the reviewer can compare the takeoff to the source without switching windows.
Mark each sampled row as pass, corrected, or escalated and keep the marks with the review note. A row that cannot be checked because the source is missing becomes an exception, not a pass. The marks make the review repeatable by another person.
Log the correction
Every correction needs a short record: the row, the original value, the corrected value, and the rule that decided the change. That record is the difference between a one-time fix and a learning event. The assistant should be able to read the note and apply the same rule to the next row.
If the note only says fixed, the next review will find the same error. Group corrections by cause rather than by person. Causes such as unclear brief, wrong revision, formula drag, and scope duplication point to different fixes.
A formula drag needs a workbook change, while an unclear brief needs a template change. Blaming the assistant rarely improves the next batch. Keep the correction log separate from the takeoff so a later reader can see the history without opening every row.
Link each entry to the row identifier and the source page. Review the log at the end of the package and look for repeated causes. The log is also the raw material for the next brief update.
Decide what escalates
Some corrections are safe for the assistant to make, and some must go to the estimator. Let the assistant fix a typo, a missing note, or a row that clearly follows an existing rule. Escalate any change that alters a quantity basis, a scope boundary, or a total used in the bid.
The estimator records the decision and the reason in the review note. Escalation should have a time limit so the estimate does not stall. Set a same-day expectation during bid week and a two-day expectation in quieter periods.
If the estimator cannot decide in time, the row stays out of the estimate rather than entering with an unresolved flag. If the deadline passes, the reviewer records the delay and the row stays out of the bid.
Sign off and release
Signoff means the reviewer accepts the sampled evidence and the open exception list. It does not mean every row was measured a second time. Record the sample size, the number of corrections, the open exceptions, and the reviewer initials.
Then release the takeoff to the estimate with the review note attached. Keep the note with the estimate version so a later change can be traced. If the estimate total moves after the review, the reviewer can see whether the move came from a corrected takeoff or from a commercial decision.
That separation is what makes the review useful in a post-bid conversation. The note is short, but it protects both the assistant and the estimator when the bid is reviewed later. Send the release note to the person who will enter the quantities into the estimate, not only to the assistant.
The note should say which version was accepted and which exceptions remain open. If the estimate is revised later, the release note stays with the earlier version. That trail prevents a corrected takeoff from being confused with an approved one.
| Check | Question | Record |
|---|---|---|
| Unit | Does the row use the brief unit? | Sampled row id |
| Arithmetic | Does quantity times unit recompute? | Original and corrected value |
| Scope | Is the row in the package only once? | Scope category |
| Signoff | Who accepts the sample and exceptions? | Reviewer initials |
