A quote is only as good as the assumptions buried inside it, and most of those assumptions never get written down. A feature gets missed on the print, a tolerance gets averaged, last year's material price gets reused, and the number that goes out the door is wrong before the customer ever opens the email. The job still runs — you just find out months later that it ran at half the margin you thought, or at a loss. To reduce quoting errors machine shop estimators have to stop treating accuracy as a matter of being careful and start treating it as a process that catches mistakes before the quote leaves the building.

This article breaks down where quoting errors actually come from — not the rare arithmetic slip, but the systematic ones that repeat across every job — and the specific controls that close each gap. The goal is a quote you can defend line by line, and a hit rate you can measure.

The five places quotes go wrong

Quoting errors are not random. They cluster in five predictable spots, and almost every mispriced job traces back to one of them.

Error type	What it looks like	Margin impact
Missed feature	A counterbore, a thread, a secondary op not seen on the print	High — unpriced work runs at a loss
Misread tolerance	±0.01 mm bore quoted as a general tolerance	High — finishing pass and scrap not costed
Stale input	Last quarter's material price or hourly rate reused	Medium — erodes quietly across many jobs
Wrong batch math	Setup amortized over the wrong quantity	Medium to high on small batches
Margin confusion	Markup applied where margin was intended	Medium — systematic underpricing

None of these are skill problems. They are reading, data, and arithmetic problems — which means they are fixable with controls, not with hiring a more experienced estimator.

Missed features: the most expensive error you cannot see

The worst quoting errors are the ones that produce a clean-looking number. A missed counterbore does not throw a warning; it just quietly omits the cost of a second operation, and the quote looks complete. Under deadline pressure, drawing review gets shortcut, and the features that get dropped are usually the small ones — a tapped hole, a chamfer that needs a separate tool, a deburr step, a heat treat callout in the notes block.

The control is a standardized feature checklist that every part runs through, regardless of who quotes it. Material and stock shape from the title block. Every hole, thread, and pocket. Every tolerance tighter than the general note. Surface finish callouts. Secondary operations — plating, anodizing, heat treat, inspection. When the same structured list captures every part, the omission becomes visible, because a blank field is obvious in a way that a missing thought is not.

Tolerances: the cost multiplier estimators average away

A drawing rarely says "this feature is expensive." It says ±0.01 mm, or specifies a position tolerance, or calls a surface finish that a roughing pass cannot hit. The cost is implied by the spec, and an estimator scanning quickly tends to average tolerances into a single mental "looks normal" — which underprices the tight features and overprices nothing to compensate.

Tight tolerances cost real money: a finishing pass, a gauged inspection loop, a possible adjust-and-rerun, and a higher scrap risk. The fix is to read tolerances as cost drivers rather than annotations. Our guide to GD&T tolerances for estimators covers which callouts add operations and how to translate a feature control frame into machine time instead of skimming past it.

Stale inputs: the error that compounds across every job

Missed features hurt one job at a time. Stale inputs hurt every job until someone notices. If your hourly rate still reflects last year's energy cost, or your material prices are the ones you typed into a spreadsheet eight months ago, every quote you send carries the same hidden discount — and because each individual quote looks reasonable, nothing flags it.

Two habits close this gap:

1. Date your inputs. Every material price and machine rate should carry the date it was set. A price with no date is a price you cannot trust.
2. Review on a schedule, not on a crisis. Re-check material costs and rebuild your machine shop hourly rate quarterly, or whenever a major input (energy, wages, a new machine) moves. Do not wait for a string of unprofitable jobs to discover the rate drifted.

A rate that is six months stale does not announce itself. It just sits underneath every quote, narrowing the margin you think you are charging.

Batch and setup: the arithmetic that flips on quantity

Setup is a fixed cost per job, and the single most common batch error is amortizing it over the wrong number — quoting the customer's annual volume but running the part in monthly releases of a tenth that size. Setup spread over 500 parts is invisible; the same setup spread over the 50 you actually make this month is not.

The discipline is to amortize setup over the release quantity you will actually machine in one run, not the headline annual number. Quote the price breaks honestly and let the customer choose: one price at the small release, a better price if they commit to the full batch up front. Getting this wrong in either direction loses money — overprice the prototype and you lose the job, underprice the small release and you eat the setup. Building the unit price up from its parts, as laid out in how to price CNC machined parts, keeps the setup contribution explicit instead of blended away.

Margin vs markup: the same word, two different prices

A 30% markup and a 30% margin are not the same number, and confusing them is a systematic underpricing error that hides in plain sight. Markup is calculated on cost; margin is calculated on price. A €100 cost marked up 30% sells for €130 and yields a 23% margin — not 30%. An estimator who thinks in markup but reports in margin is quietly leaving several points on every job.

Pick one and apply it consistently. If you want a 30% margin, the formula is price = cost ÷ (1 − 0.30) = cost ÷ 0.70, which on a €100 cost gives €142.86. The margin vs markup breakdown shows why the gap widens as your target margin climbs, and why "we add 30%" is an ambiguous instruction that two estimators will price differently.

Make accuracy a checklist, not a personality trait

The shops with the most consistent quotes are not the ones with the most careful people. They are the ones where accuracy does not depend on a person being careful. The difference is a pre-send checklist that every quote clears before it goes out:

• Every feature on the print captured and costed — no blank fields.
• Tolerances tighter than the general note priced as operations, not averaged.
• Material price and hourly rate dated within the review window.
• Setup amortized over the real release quantity.
• Margin (not markup) applied at the target percentage.

Five checks, two minutes, and the five most expensive error types are caught before the customer sees the number. A consistent quote template that forces each of these fields turns the checklist from a habit you have to remember into a structure you cannot skip.

From drawing to a quote that catches its own errors

Most of these errors share one root cause: the drawing gets read by a tired human under deadline, and reading is where features get missed and tolerances get averaged. QuoteBuddy attacks the problem at the source — it reads the technical drawing with AI drawing interpretation and surfaces the material, dimensions, holes, threads, tolerance classes, and implied operations as a structured list, so the estimator confirms a complete picture instead of scanning for what might be missing.

From there the cost engine builds the unit price from your machine rates, dated material costs, setup times, and target margin — the same way every time, with margin and markup kept straight by the math rather than by memory. The quote that comes out is one where a missed feature shows as a blank, a stale rate is flagged by its date, and the price is the price the job was actually estimated to carry.

Start a 30-day trial and run a handful of real drawings through it. Compare the quote it builds to the one you would have produced by feel, and see how many of the five error types it catches before they reach the customer.