Every estimate starts from scratch

A Thursday evening a long time ago, in the company I owned before this one. I was parked outside a customer's building with the dome light on, writing a number on the back of a work order. The man wanted a price before I drove away, so I gave him one. He took it in the doorway.

Eleven weeks later that job closed out about four thousand dollars light. I went looking for how I had built the number and there was nothing to look at. A figure I had said out loud in a parking lot. By then I could not have told you what I thought was inside it.

I filed that away as a bad job. Then I got a little sharper about that kind of work and carried on handing people numbers out of the truck for another two years.

Twenty-six men and one man who can price a job

A structural and miscellaneous metals shop on the east side of Lubbock. Two bays, a press brake, a plasma table, twenty-six on the payroll. Stairs, handrail, mezzanines, skid frames for the oil patch. A bit over five million a year, and it had grown in each of the last six.

Ray built the estimates. All of them. He did his take-offs at the kitchen table four nights a week with the drawings weighted down at the corners and the television going in the next room. His foreman could price small repair work, and the front counter could quote handrail by the foot. On anything larger the three of them never landed within ten percent of each other, and each could defend his number.

None of this looked like trouble from the street. The bays were full, the trucks were newer than the ones across town, and the work that left the shop was clean. Some of it will sound like your place:

The last four bids you sent, only you could have produced. Three weeks in the hospital and quoting stops in your company.

There is no page anywhere that says how an estimate gets built here. The method is in your head, and a fair amount of it is in your hands.

Consumables, gas, grinding wheels, the forklift, delivery, the trip back for the punch list. All of it happens on every job and none of it is in any estimate.

Profit is whatever survives after you decide what this customer will take, so it is the first thing to move when you want the work.

Your books carry one line for labor and one for materials, so a job that goes badly reads as a job that went badly. Where it went is not in there.

A price can leave your building without a second person seeing it, at any size, on any day.

The money was always about to loosen up

Ray had a reason and it was a decent one. Money was tight. Steel gets paid for well before the progress billing lands, payroll runs every Friday no matter what, and a fabricator that size carries more in work-in-process than most owners care to think about. So he bid to keep the bays full and told himself the rest would get built properly the day the money loosened up. After the tank farm job. After the note on the plasma table came off.

What owners buy in that state is money. A bigger line of credit, an equipment note restructured, sometimes factoring on receivables at a rate they would laugh at anywhere else. It works, in the sense that Friday keeps happening, and it removes the only signal that anything is wrong.

Cash sits downstream of the number Ray writes at night. The estimate is where a company decides what it will be paid and what it gets to keep, and when that decision runs short the volume makes it worse. The error rides on every job, so winning more of them buys more of the error.

The next failure is the dull one. A man accepts that estimating needs a method, so he builds one. A spreadsheet with a markup at the bottom. That is a real improvement and it stops paying about a month in. A template makes the same estimate faster. What makes the next estimate better is the dull hour after a job closes, when somebody sets actual cost beside estimated cost and works out where the difference came from. Nobody owns that hour. It is never urgent, and it is the whole mechanism by which a shop gets good at pricing its own work.

What it costs to build the same quote twice

We went through the bids Ray had sent in the previous three weeks. Three a week, near enough, were jobs the shop had effectively built before. A stair run to a different height. A skid frame with a different pump on it. Same details, same welds, same finish, and every one started from a blank page, because nothing survived the last one except an invoice.

Call it two man-hours to rebuild each, Ray and the foreman between them, at a blended thirty-eight dollars an hour.

That is the cheap end, and it is the part you can watch happen. Four jobs closed under water last year and came to sixty-three thousand between them. Ray could not tell me what those four had in common. Estimates sit in one folder and job costs sit in the accounting system under headings that were never built to match.

Every bid carried a markup, twenty percent, and it moved whenever a job mattered enough. Overhead came out of that markup. Profit came out of what the markup had left afterward, which made profit the last thing in the building to get paid and the first thing handed away at a counter.

Ray can sell. His men can build anything that comes through the door. What the company could not do was pay for its own growth, and that ceiling had been set months earlier in a number nobody checked.

Fifteen days and a spiral notebook

Buy a wire-bound notebook and leave it open where the numbers get made. The estimating desk, the plotter, the kitchen table. For the next fifteen working days, every price that leaves the building gets one line in it before it goes.

Date. Customer. What the job was. The price. Who built it. Roughly how long it took to build. Then one word at the end for how the number came about.

That last word carries the whole exercise, so allow three and no others. TAKEOFF, when the number came off the drawings with a written bill of material and labor priced against it. HISTORY, when it came from a job the shop has done before. FEEL, when it came from a man who knows steel and had to say something that afternoon.

The person who built the estimate writes the line, before the quote goes out. Not you, on Friday, from memory. A line is twenty seconds of work, and if it is taking longer you have made it too complicated.

On the fifteenth day, take the notebook somewhere quiet and sort every line into those three stacks. Add up the dollars quoted in each, then the hours spent in each. Owners get a shock in one direction or the other. Either FEEL is carrying far more of your revenue than you would guess, or TAKEOFF is eating nights the work it wins does not justify.

Then, with the notebook still open, pull the last ten jobs that closed and write estimated cost against actual cost on one sheet, sorted by the size of the gap. Ten jobs, one afternoon. You are looking for whether the misses are scattered or whether they lean one way. Scattered misses mean your judgment is running hot and cold. A lean means the same line item is missing from every estimate you write, and it usually turns out to be fit-up and grinding, or the crane you rent on every third job and price on none of them.

Set one number while you are at it. Above some size, no bid leaves the building until a second person has read it. Pick a figure you can say out loud today, twenty thousand or fifty, whatever tightens your stomach a little. The second reader only has to be somebody who did not spend four hours on it.

The bids that stop coming back to your desk

Six or eight weeks along, the first thing you notice is quiet. The nine o'clock phone call about what to put on a job stops. The Monday spent rebuilding a quote somebody in the building already built stops. Your foreman stops carrying a finished job's paperwork into your office to tell you it went sideways, because the pair of you looked at it while it was still running.

Then something smaller. You start recognising jobs again. A drawing comes across and somebody says we built that one already, and the file comes out, and a number that used to cost you an evening now costs you forty minutes.

Ray still does take-offs at night. Fewer of them, and the shop keeps them now. An estimate is a promise about what your company can do and what it will accept for doing it, made before anyone strikes an arc. A shop that cannot say where its numbers come from is guessing at its own future in twenty thousand dollar pieces. His men could always build it. The number was the only thing in that shop nobody built.

Estimating Process is eight statements in the Business Check-in, and this article covers most of them. The rest of that section asks about the things a busy shop stops noticing, and the check-in is free and runs about ten minutes.