Finish this job to standard, price only the change orders that are genuinely changes, and eat your own miss. Then put the estimate and the actual cost side by side at line level to find out which of three things happened: a line was priced too low, a quantity was wrong, or a line was never there at all.
Written by JD Davenport, who still sells for a living. Updated 4 November 2026.
The first is this job, which is mostly a decision about how to finish it. The second is whether your estimating is wrong, which is the expensive one, and you cannot answer it from a feeling.
Finish it to standard. A job you lose money on and do properly costs you the margin once. A job you lose money on and rush costs you the margin, the callback, the review and the referrals that customer was going to send you.
Do not go looking for the loss in change orders you cannot justify. A customer who feels nickeled after a bad bid tells people, and your cost to acquire the next three jobs just went up by more than this one lost.
Price the change orders that are genuinely changes. Eat the ones that were your miss. Write down which was which while you still remember, because that note is the only honest record you will have in three months.
Put the estimate and the actual cost side by side at line level. Not the totals. The totals only tell you the size of the hole, and you already know that.
Three things can be true and they have different fixes.
A line was priced too low. You had the right quantity at the wrong rate. This is the easiest to fix and the least common.
A quantity was wrong. You priced eighteen squares and it took twenty-two. Measurement or takeoff, and it will repeat on every job of that shape until the takeoff changes.
A line was never there at all. Disposal. A second trip. Steep charge. Permit. Rental. Plywood you assumed you would not need. This is the one that gets people, because nothing on the estimate was wrong. The estimate was incomplete, and an incomplete estimate looks exactly like a correct one right up until the invoices land.
A rate you priced wrong, you will notice. A line you never priced at all, you will not notice, because there is nothing on the page to check.
That is a structural problem and it has a structural fix. The fix is not more care at the kitchen table. It is that the estimate cannot be produced without every line being addressed, including the ones that usually come back as zero. A line marked not applicable is a decision. A line that is absent is an accident, and the two are indistinguishable once the job is sold.
If your estimate is a blank page you fill in, you will keep omitting things under time pressure in a customer's driveway. If it is a list you must answer, the omission has no route in.
One underbid job is a Tuesday. The question is whether it is the only one, and the only way to know is to run the same comparison against your last ten finished jobs.
Sort them by the gap between quoted and actual. If one job is far out and the rest are tight, you had a bad day and the system is fine. If the same line is short on six of ten, you do not have an underbid job. You have an estimating defect that has been billing you quietly for however long it has been there.
That is worth a day of work to find, and it is the cheapest day you will spend this year.
The line-level comparison, the tests that surface which lines keep coming up short, and the workbook are The Job Cost Standard. The estimate that cannot be submitted with a line unanswered is The Appointment Standard. More answers on the questions page.