1. Normalize every order into production inputs
For each job record good quantity, stitches per piece, cap or flat workflow, same-design batch size, handling time, due date and whether the job is actually ready. Piece count alone is weak because 100 five-thousand-stitch polos and 100 thirty-thousand-stitch jacket backs do not consume the same machine time.
Use measured minutes per 1,000 stitches when possible. If you do not have them, time several representative runs with the Shop Calibration Calculator and keep flat/cap assumptions separate when the workflows behave differently.
2. Separate readiness from capacity
Artwork approval, blank inventory and customer decisions are upstream constraints. A machine can have open time while production still cannot start. Keep blocked jobs out of the ready queue and make their blocker visible. This stops planners from treating “orders sold” as “work available to stitch.”
3. Build a weekly capacity budget
Start with available machine hours and operator hours. Convert the ready workload into machine minutes and handling minutes, then compare both. The smaller capacity wins because production cannot exceed the bottleneck.
For synchronized multi-head equipment, include batch utilization. A six-head machine cannot use six heads on a three-piece same-design batch. For independent heads, account for parallel jobs instead. The Queue Lab models both architectures against the same set of jobs.
A simple capacity budget
| Bucket | What belongs there |
|---|---|
| Committed production | Ready customer work already promised. |
| Known rework | Replacement pieces and quality recovery already identified. |
| Reserve | Capacity intentionally left unsold for variability, rushes and maintenance. |
| Uncommitted capacity | Space still safe to sell after the other buckets. |
4. Protect a buffer before the week fills
If the plan consumes every sellable hour, it has no way to absorb ordinary variation. Choose a reserve based on your own history: how often rush jobs arrive, how much rework occurs, whether caps are unpredictable, how concentrated machine risk is and how seasonal the demand pattern becomes.
The capacity-buffer guide explains how to choose and review this reserve. The Growth & Peak Planner shows when growing demand consumes it.
5. Control backlog as a flow, not a pile
Backlog only clears when daily good capacity exceeds new work entering the ready queue. If 90 pieces leave production and 100 ready pieces arrive, the backlog grows by 10 per day even though everyone is busy.
Use the Backlog Clearance Planner to calculate daily capacity, net clearance and estimated days to normal. It also compares simple recovery levers such as extra operator capacity, extra heads and an extra production hour so you can see which lever actually reaches the bottleneck.
6. Use a daily operating loop
- Morning: confirm which jobs are ready, what became blocked and any new rush work.
- Before loading: sequence by deadline, then by compatible setup where deadlines allow.
- Midday: check whether machine or operator wait time is accumulating.
- End of day: record good pieces completed, backlog change and any run-time variance.
- Weekly: update measured assumptions only when repeated evidence supports the change.
Worked example: busy is not the same as recovering
Suppose a shop begins Monday with 300 ready pieces in backlog. Its measured good capacity after reserve is 120 pieces per day, while 80 new ready pieces arrive each day. Net clearance is only 40 pieces per day, so the original backlog needs about 7.5 production days to disappear. If new arrivals rise to 125 per day, the backlog cannot clear at all without more capacity or lower demand.
This distinction helps avoid a common mistake: adding overtime for one night when the underlying arrival rate still exceeds sustainable output. Fix the flow first, then decide whether overtime, staffing, process improvement or equipment is the appropriate response.
7. Keep a small planning history
You do not need an ERP to learn from actual production. Keep a weekly record of planned good pieces, actual good pieces, overtime hours, rework, backlog at the start and end of the week, and the largest cause of variance. Over several weeks this becomes a shop-specific evidence base for capacity decisions.
When a number repeatedly misses in the same direction, update the assumption. When one unusual job creates a single miss, preserve the normal baseline and treat that job as an exception. This prevents overfitting the model to the last bad day.