In a small job shop, yield loss rarely shows up as one dramatic event. It leaks out a few pieces at a time: extra material consumed during setup, first-off parts adjusted into spec after several tries, parts that drift during a short run, and leftover material or unfinished pieces that get written off at cleanup. On long production runs, those losses can be diluted. On short runs, they can quietly erase the margin you thought you had.
The problem is that most shops track scrap in one bucket. That hides the pattern. If you want to protect high-mix margins, you need to separate yield loss into when it happens: startup, in-process, and end-of-run. Once you tag those losses by changeover, machine, part family, and operator, the expensive jobs become obvious. You can then fix quoting, setup methods, inspection timing, and run-size decisions with data instead of hunches.
Why short runs create disproportionate yield loss
Short-run work carries fixed effort that does not shrink just because the lot size is small. Every changeover has setup labor, trial pieces, first-article verification, material handling, and closeout time. If a job runs 25 parts instead of 2,500, those losses are spread over far fewer good parts.
That is why a shop can have acceptable overall scrap percentage but still lose money on high-mix work. A 2-piece startup loss on a 20-piece order is a 10% hit before the run really starts. Add one or two pieces lost to mid-run adjustment and a small quantity left as unusable remnant at cleanup, and the job margin can collapse.
This is especially common when shops schedule frequent switches across materials, tooling, fixtures, or revisions. If your environment is high-mix low-volume, measuring by job alone is not enough. You need to measure by changeover event.
For broader context on digital shop-floor visibility, see this manufacturing execution system software guide and the NIST manufacturing resource at NIST.
The three loss buckets to track on every short run
1. Startup scrap
Startup scrap is everything consumed from setup start until the first approved good piece. Depending on your process, that can include:
- Trial parts cut to dial in offsets
- Material used for warm-up or machine stabilization
- First-article pieces rejected before approval
- Labor spent producing nonconforming setup pieces
- Extra inspection time required to reach first acceptable output
Many shops record setup time but not setup scrap. That is a mistake. On short runs, startup scrap is often the largest yield loss category.
2. In-process loss, including first-off drift
In-process loss happens after the first approved part and before the planned run quantity is complete. For short runs, one specific subtype matters: first-off drift. This is when the first approved part is in spec, but the next few pieces move out of tolerance as the process settles, tooling seats, temperatures shift, or offsets are refined.
Track in-process loss separately from startup because the root causes are different. Startup issues point to setup method, fixturing, work instructions, or first-article release timing. In-process drift points to process stability, tool wear, compensation practices, or inspection frequency.
3. Cleanup loss
Cleanup loss is what gets consumed or stranded at the end of the job because the run stops. Examples include:
- Leftover cut lengths too short for reuse
- Mixed or unlabelled partials that cannot return cleanly to inventory
- WIP pieces abandoned because the order closed short
- Purge, flush, or washout material
- Labor spent sorting, counting, and disposing of residual material
Cleanup loss is easy to ignore because it happens after the “real work” is done. But in high-mix shops, end-of-run waste can be repeated dozens of times per week.
What to capture at the changeover level
If you only measure total scrap by month, you will not find the source of margin erosion. Create a simple event-level record for every short-run changeover. At minimum, capture:
| Field | Why it matters |
|---|---|
| Work order / job number | Links loss to quote, customer, and actual run |
| Part number and part family | Shows repeat patterns across similar jobs |
| Machine / work center | Separates process-specific issues |
| Previous job to next job | Reveals bad changeover pairings |
| Material type and gauge / size | Identifies material-sensitive startup loss |
| Setup start and first good timestamp | Measures startup duration and first-piece approval delay |
| Startup scrap quantity and cost | Quantifies setup-related yield loss |
| In-process scrap quantity and reason | Shows drift, handling, and process control issues |
| Cleanup loss quantity and cost | Captures remnant and closeout waste |
| Operator and shift | Useful for training and standard work analysis |
| Reason code | Makes trend analysis possible |
Do not overcomplicate the first version. The goal is to make loss visible. If your team can consistently classify loss into three buckets with a small set of reason codes, you will learn a lot very quickly.
Use reason codes that point to action
Generic scrap codes like “bad part” or “setup issue” do not help. Use codes that reflect causes you can fix. A practical starter list might include:
- Offset adjustment after first-off
- Fixture alignment
- Tool seating or tool wear
- Program revision mismatch
- First-article approval delay
- Material bow, thickness, or lot variation
- Operator handoff / instruction gap
- End-of-run remnant below reusable size
- Uncounted partials at closeout
- Purge / flush required by product change
If revision confusion affects startup and first-off quality, this article on manufacturing ECO release tracking is relevant. If first-article timing is delaying setup completion, see manufacturing first-article approval tracking.
How to calculate true short-run yield loss
Most shops stop at material scrap cost. That understates the problem. True short-run yield loss should include:
- Material loss: raw material consumed by startup, drift, and cleanup waste.
- Labor loss: setup, run, inspection, sorting, and closeout labor tied to nonconforming or stranded output.
- Machine time loss: time spent producing non-sellable parts or repeating setup adjustments.
- Overhead impact: if you burden labor or machine rates internally, apply the same logic here.
A simple formula is:
Total short-run yield loss = startup loss + in-process loss + cleanup loss
And for each bucket:
Bucket loss = material cost + labor cost + machine cost
Example, purely illustrative:
- Startup: 3 pieces scrapped at $18 material each, plus 20 minutes operator time and 10 minutes inspection time
- In-process: 2 pieces lost during first-off drift, plus 12 minutes of machine time for adjustments
- Cleanup: $22 of unusable remnant plus 8 minutes of closeout labor
On a 30-piece order, that may be the difference between a profitable job and a job that only looked profitable on paper. If you want a quick baseline for material impact, use the free scrap cost calculator.
Focus on changeover pairs and part families, not just single jobs
One of the most useful views is not “Which jobs had scrap?” but “Which transitions create loss?” For example:
- Stainless to aluminum may require more cleanup and contamination control
- Thick gauge to thin gauge may create setup rework on the same brake or press
- Complex turned parts with tight first-off requirements may repeatedly drift in the first five pieces
- Short printed packaging runs may always leave excess material at end-of-roll
When you analyze by previous job to next job, patterns emerge that quoting by part number alone misses. Likewise, part-family analysis can show that a whole class of work is underpriced because it reliably carries high startup loss.
This also ties into scheduling. If certain sequence changes create high cleanup loss, group compatible jobs where possible. For ideas on that, see high-mix low-volume production scheduling.
Build a practical tracking workflow on the shop floor
Step 1: Define the start and stop points clearly
Everyone must use the same definitions:
- Startup: setup start to first approved good part
- In-process: after first approval to last planned piece completed
- Cleanup: after run completion until residual material and WIP are dispositioned
Without those boundaries, data will be inconsistent.
Step 2: Record quantities immediately
Do not wait until end of shift. Operators or leads should log scrap and residuals at the moment they occur. Real-time entry is more accurate than backfilled paperwork. If you are trying to collect better floor data without expensive automation, this guide on real-time shop-floor data without IoT is a good place to start.
Step 3: Require first-good confirmation
A timestamp for first approved part is essential. Without it, setup time and startup loss get mixed into run time. That makes machine performance look worse and hides quality loss at the start of the job.
Step 4: Capture partial completions and leftovers cleanly
Cleanup loss is often hidden by poor closeout discipline. If quantities move in batches, split across containers, or remain partially complete, use a consistent closeout process. This is where good WIP tracking matters; see manufacturing partial completion tracking.
Step 5: Review weekly by the right dimensions
At a minimum, review yield loss weekly by:
- Changeover pair
- Part family
- Machine / work center
- Operator / shift
- Customer or quote family
Monthly summaries are useful for trends, but weekly review is what drives action.
What to do with the data once you have it
Fix bad quotes
If a part family always consumes two startup pieces and 15 minutes of extra first-off adjustment, that cost belongs in the quote. Many small shops underquote short runs because they assume historical average scrap from longer jobs. Separate short-run yield data lets you quote with real setup and waste factors.
Set minimum lot sizes or setup charges
Some jobs are simply too small to run at the current price. Your data may justify a minimum order quantity, a setup charge, or a material remnant policy for certain product types.
Standardize setup for repeat families
If startup scrap clusters around specific machines or operators, standard work may be missing. Improve setup sheets, fixture presets, offset checklists, tool staging, and first-piece verification steps.
Reduce approval delays
When the first good part sits waiting for inspection or signoff, setup stretches and drift risk rises. Digital inspection workflows can help; see the digital quality inspection checklist guide.
Improve scheduling sequences
If cleanup loss rises on certain product switches, sequence compatible jobs together. In high-mix shops, smarter sequencing can reduce both waste and schedule disruption.
Common mistakes to avoid
- Lumping all scrap together: you lose the ability to see when and why the waste occurs.
- Ignoring labor and machine time: material-only scrap costing understates margin erosion.
- Skipping cleanup loss: remnants and closeout time matter on short runs.
- Using too many reason codes: start simple so people actually use them.
- Reviewing only by total dollars: low-cost parts may still consume large percentages of margin.
- Not connecting to quoting and scheduling: measurement without process change will not improve profit.
A simple scorecard for small job shops
If you want one page your team can review each week, include these measures:
- Startup scrap pieces per changeover
- First-good time by machine and part family
- In-process scrap in first 10 pieces after approval
- Cleanup loss dollars per work order
- Total yield loss dollars per good part shipped
- Top 10 changeover pairs by loss
- Top 10 part families by short-run margin erosion
This scorecard gives supervisors and owners a much clearer view than monthly scrap percentage alone. It shows where high-mix complexity is costing money and where to act first.
Conclusion
Short runs do not just challenge scheduling; they can quietly consume margin through startup scrap, first-off drift, and cleanup loss that never gets measured separately. When you track yield loss by changeover event and split it into startup, in-process, and end-of-run buckets, the hidden cost of high-mix work becomes visible. That lets you quote better, schedule smarter, improve setups, and stop subsidizing the wrong jobs.
If you want a practical way to capture shop-floor production, scrap, and closeout data in real time, start a free FactoryOS trial.
