Research

Scrap Rate Benchmarks for Metal Fabricators: What 1.4% Really Costs

The FMA Financial Ratios survey reports scrap and rework averaging 1.4% of sales for US metal fabricators. For a $5M revenue shop, that's $70,000/year. Top-quartile shops keep it below 1.0%. AI-native ERP routinely cuts scrap 20 to 30% by surfacing shop-floor signals in real time.

Published May 2026 Author Tangle Research Audience Metal fabrication operators and leaders
From the parent brief. This article expands one of the five levers in Five Levers, One Year — Tangle Research's executive primer on AI-native ERP ROI in metal fabrication.

1. What "scrap and rework" actually means

Different shops count differently. The FMA definition — used in the 1.4%-of-sales benchmark — captures both pure scrap (parts that hit the skip) and rework (parts that fail inspection and have to be reprocessed). Both consume material, machine time, and operator time. Both drop straight to the bottom line.

Most owners under-count scrap because the visible portion — the parts that are obviously bad at final inspection — is only the tip. Parts that fail in-process and get re-cut without paperwork are usually invisible. So is the indirect scrap created by re-runs that push other jobs late, which then have their own quality compromises under expedite pressure.

When a shop measures scrap honestly for the first time, the number is almost always higher than the owner expected.

2. Where scrap actually comes from in a custom shop

Three patterns produce most scrap in custom fabrication. Programs that have not been validated against the actual part revision. Material variability the operator did not see until a part was already cut. Setup errors on small batches where the cost of a thorough first-article check is hard to justify against batch size.

All three share a common root cause: the signal that something is wrong arrives too late. The misfed sheet finishes its run before anyone notices the orientation. The wrong tool offset persists until twenty parts later. The tolerance drift on a worn bend die builds up over a week. By the time the data reaches the supervisor, the metal is already in the skip.

3. What AI-native ERP changes

AI-native ERP collects shop-floor signals continuously rather than at shift-end. Machine telemetry. Lightweight operator confirmations. In-process measurement. All feeding a system that can flag deviations in real time. The flag does not have to be sophisticated. "This cycle ran 12% slower than the last 50 cycles of this part" is enough to interrupt a problem before it becomes scrap.

Tighter job costing also exposes the chronic loss-makers — the customers, parts or processes that have been quietly destroying margin for years. Once visible, they can be repriced, redesigned or declined. This is a structural scrap reduction that does not show up in published case studies but matters more over time than the in-process catches.

4. The math

Annual scrap cost = annual revenue × scrap % of sales. The conservative case applies a 20% reduction. The realistic case 30%. Both sit at or below the published range for AI-native ERP deployments.

Scrap & rework impact — $5M shop at 2% current rate
ScenarioScrap reductionAnnual scrap costAnnual saving
Baseline$100,000
Conservative20%$80,000$20,000
Realistic30%$70,000$30,000

5. How to measure your shop honestly

Pick a single quarter. Track three numbers. Parts scrapped (count and material cost). Rework hours (operator time at fully-loaded rate). Re-runs caused by quality issues (machine time displaced from other revenue work).

Add them. Divide by revenue. The resulting percentage is the honest scrap rate. It will be higher than the management report number. The honest version is what AI-native ERP can move.

When a shop measures scrap honestly for the first time, the number is almost always higher than the owner expected.

Run the model with your own numbers

Three to five minutes. Five inputs. Same framework, applied to your shop.

Open the ROI calculator

· Frequently asked questions

What is the average scrap rate for metal fabricators?

The FMA Financial Ratios & Operational Benchmarking Survey reports scrap and rework averaging 1.4% of sales for US metal fabricators. Top-quartile shops keep this below 1.0%. Automotive-supplier targets sit below 0.5%.

How does AI-native ERP reduce scrap in real terms?

By collecting shop-floor signals in real time and surfacing deviations in minutes rather than at final inspection. The catches happen mid-run, not after the fact. Tighter job costing also exposes the chronic loss-maker parts that produce scrap structurally — allowing the shop to reprice, redesign or decline them.

Is the 20 to 30% scrap reduction figure realistic?

Published ERP case data ranges 25 to 40% scrap reduction with disciplined implementation. The 20 to 30% range in this model sits at the lower end deliberately, to keep the headline number defensible.

What if my shop does not currently measure scrap?

Use the FMA 1.4% of sales average as a placeholder. The real number is usually higher. Spending a single quarter measuring honestly is the highest-leverage diagnostic an operations leader can run.

· Sources

  1. FMA Financial Ratios & Operational Benchmarking Survey — The 1.4% of sales industry-average scrap benchmark.
  2. RMDB / User Solutions — Scrap Rate Reduction guide — Top-quartile and industry-specific scrap targets.
  3. The Fabricator — Financial survey dives deep into metal fabrication industry — Industry context and survey methodology.