Build Mistakes

Costly Estimating and Execution Mistakes on Capital Equipment Builds

A troubleshooting guide to the errors that quietly erode margin on engineered-to-order machines, from omitted commissioning hours to backlog counted at list price.

The single most expensive mistake in machine building is quoting hardware cost and treating engineering as a rounding error. Symptom: a machine that costed at 22 percent margin ships at 6 percent. Root cause is non-recurring engineering absorbed into overhead instead of the quote. On a 40 hour standard build, controls and mechanical design routinely run 180 to 400 hours at a loaded 95 to 140 dollars an hour, so 30k of design vanishes if you skip Engineering Hours per Machine. Fix: capture design hours as a line item and recover them across the order quantity, not as a hidden burden.

Second common failure is unit confusion between hours, days, and shifts on FAT and SAT planning. Symptom: acceptance runs three weeks longer than the schedule promised. Root cause: someone entered a test protocol with 120 checks assuming an 8 hour clearance rate, when the measured completion rate is closer to 4 to 6 checks per hour once retest and punch-list items are counted. Run FAT Workload and SAT Workload with an honest 15 to 30 percent retest allowance. A protocol that looks like 15 hours on paper is 26 to 34 hours in the test bay.

Third: pricing custom options at standard-machine margin. Symptom: your most configured orders are your least profitable. Root cause is that a bolt-on option carries mechanical, panel, and controls content that scales faster than the base price. A configuration that adds 18 percent to sell price can add 35 to 50 percent to engineering hours. Score departure from the base build first, then price the delta with Custom Option Cost so complexity flows into the quote instead of eating margin. Never quote a heavily optioned unit off the base configuration sheet.

Fourth: setting warranty reserve as a flat percentage with no claim history. Symptom: the reserve is fine on average but two field failures wipe out a program's margin. Root cause: a flat 1.5 percent of sell price ignores that first-article and new-configuration machines fail 2 to 4 times more often than mature designs. Tie Warranty Reserve to a real failure and claim rate per installed dollar, and carry a higher rate, often 3 to 5 percent, on the first two or three units of a new platform before the field data justifies dropping it.

Fifth: omitting commissioning and site travel entirely from the estimate. Symptom: the project is profitable at FAT sign-off and underwater after startup. Root cause: field time is scoped as if it mirrors the shop, but site work runs 1.4 to 1.8 times slower and adds per-diem, travel, and idle days waiting on customer utilities. A two-technician startup budgeted at 40 hours often lands at 70 to 90 billable-equivalent hours. Use Commissioning Cost and Field Install Cost with a realistic travel and standby line, not just wrench time.

Sixth: valuing backlog at list price and calling it cash. Symptom: the order book looks healthy but working capital is tight and a slipped delivery triggers a penalty. Root cause: Machine Backlog Value counted at sell price ignores unearned margin, long-lead exposure, and liquidated-damages clauses. Value backlog at expected build cost plus recognized margin, and flag any line where a single casting or drive with a 20 to 30 week lead time can move the whole ship date. One long-lead part slipping four weeks can cascade into a 6 to 8 week delivery miss.

Seventh: assembly labor loaded from an ideal routing that ignores learning curve and rework. Symptom: the first unit of a new build blows its labor budget by 40 percent. Root cause: standard hours are set from the third or fourth unit, not the first, and no scrap or re-do allowance is carried. Apply an 80 to 90 percent learning curve so unit one carries 25 to 60 percent more assembly hours than steady state, and put a 5 to 10 percent rework line in Assembly Labor Load. Estimating the whole run at mature hours guarantees an early-unit overrun.

The cross-cutting fix is a post-mortem that compares quoted content to actual on every completed machine. Pull engineering, assembly, FAT, SAT, and commissioning hours against the estimate and compute the variance per phase, not just at the total. Most shops find leakage clusters in one or two stages: usually controls hours and site commissioning. Feed those actuals back into your next Engineering Hours per Machine and Commissioning Cost inputs. A shop that closes this loop typically cuts estimate variance from plus or minus 25 percent down to under 10 percent within a year.

Published 2026-07-01.