Production planning in a small workshop has a reputation for being either unnecessary or impossible. Unnecessary because everyone can see the whole operation from where they stand; impossible because it changes daily anyway.
Both are wrong for the same reason. The plan is not a schedule you obey — it is the arithmetic that tells you whether what you have promised is physically achievable, before you promise it. That calculation takes about an hour a week and prevents most of what goes wrong.
Three things a plan reconciles
Every production plan balances the same three constraints, and a plan that ignores any one of them will fail:
- Demand — confirmed orders, plus a forecast for what you make to stock.
- Capacity — the productive hours you genuinely have, at the resource that limits you.
- Materials — components in stock or arriving in time, per the bill of materials.
Most small workshops plan against demand alone. The result is a schedule that is fine on paper and collides with reality in week two, usually at a curing step or a single machine nobody was counting.
What your capacity actually is
Start with staffed hours, then subtract everything that is not production. Three people at 37.5 hours is 112.5 staffed hours a week, and it is not 112.5 hours of output.
| Deduction | Typical share | Hours |
|---|---|---|
| Staffed hours | 112.5 | |
| Breaks and shift start-up | 8% | -9.0 |
| Cleaning and maintenance | 4% | -4.5 |
| Admin, meetings, interruptions | 3% | -3.4 |
| Available productive hours | 95.6 |
Measure these percentages once rather than assuming them. Most workshops that time themselves honestly find non-productive time is higher than they expected, not lower.
Plan to 85%, not 100%. A schedule loaded to full capacity has no absorption for a sick day, a machine fault, or a rush order — and those are not exceptions, they are the normal texture of a week. Planning to roughly 85% of available hours is what makes a plan survive contact with a Tuesday.
Load the week and check it
With 95.6 available hours, and a target load of about 81 hours, next week's demand looks like this:
| Product | Units | Run time each | Batches | Setup each | Total hours |
|---|---|---|---|---|---|
| A | 120 | 12 min | 2 | 30 min | 25.0 |
| B | 60 | 22 min | 1 | 45 min | 22.8 |
| C | 200 | 9 min | 2 | 20 min | 30.7 |
| Total | 78.5 |
78.5 hours against 95.6 available is 82% loaded — a plan with enough slack to absorb an ordinary week. Had it come out at 104 hours, that is not a scheduling problem to be solved by working harder; it is a commitment problem, and the time to discover it is before the delivery dates are confirmed.
Find the constraint
Labour hours are rarely the real limit. Somewhere in the process there is one resource with less capacity than everything around it, and that resource sets the output of the entire workshop.
Suppose products A and C both need a curing oven that holds 40 units and runs a 4-hour cycle, two cycles a day, five days a week — 400 units of oven capacity. Demand for A and C is 320 units, 80% loaded. Tight but workable. Add a rush order of 100 units of A and the labour plan still fits comfortably while the oven is at 105% and physically cannot deliver.
Once you know where the constraint is, a few rules follow directly:
- An hour lost at the bottleneck is an hour lost for the business. An hour lost anywhere else costs nothing, because that resource had slack.
- Never starve it. Keep a small buffer of work in front of the constraint so it is never idle waiting for the previous step.
- Inspect before it, not after. Processing a defective unit through the constraint consumes capacity you can never recover.
- Schedule the whole workshop to the constraint's rate. Producing faster upstream just piles up work in progress.
The constraint moves as your mix changes. Re-check it when you add a product or change volumes significantly, rather than assuming last year's bottleneck is still this year's.
Sizing batches
Batch size is a trade-off between setup and holding. Small batches mean setup consumes a large share of capacity; large batches mean cash and space tied up in stock you will not sell for months, plus longer lead times because everything else waits behind the run.
Setup share of a run = setup time / (setup time + units x run time per unit)
For product B — 45 minutes of setup and 22 minutes a unit — a batch of 10 spends 17% of its time on setup. A batch of 60 spends 3.3%. That is the pressure toward larger runs.
The pressure the other way is holding cost, and the same square root that gives economic order quantity applies:
Batch size = √( 2 x annual demand x setup cost / annual holding cost per unit )
With annual demand of 3,000 units, a setup costing EUR 12.30 in consumed capacity, and EUR 3.10 to hold a unit for a year, the batch works out at about 154 units. As with EOQ, the answer is insensitive — being 25% out raises total cost by roughly 2.5% — so use it to find the right scale, then round to something practical like a full tray, a full oven load, or a whole pack of material.
The high-leverage move is usually not optimising batch size but attacking setup time itself. Halving setup on product B cuts the economic batch to about 109 units and shortens lead times, without any cost increase. Preparing tooling in advance, standardising fixtures, and sequencing similar colours or flavours together are unglamorous and they change the arithmetic permanently.
Make to stock or make to order
| Make to stock | Make to order | |
|---|---|---|
| Customer waits | No | Yes |
| Cash tied up | High | Low |
| Obsolescence risk | Yours | Minimal |
| Batch efficiency | Good | Poor for one-offs |
| Best for | Steady sellers, short customer patience | Variants, personalisation, high value |
Most workshops need both, and the useful structure is a hybrid: make intermediates to stock, assemble to order. Blend the wax, dye the yarn, cut the panels in efficient batches — then finish to the specific order. The customer waits only for the assembly step, while you keep batch efficiency where the setup cost actually is. This is a direct argument for structuring intermediates properly in your bills of materials.
Work in progress hides problems
Half-finished work feels like progress. It is not: it is cash converted into something that cannot be sold, and it conceals the flow problems that caused it.
Production lead time = work in progress / throughput rate
This relationship — Little's Law — is unforgiving. Doubling work in progress doubles lead time at unchanged throughput. A workshop drowning in part-built product does not have a productivity problem; it has released more work than the constraint can absorb. Releasing less work makes everything move faster, which is counterintuitive enough that most people have to see it to believe it.
A weekly rhythm that fits in an hour
- Confirm demand. Orders due, plus replenishment for make-to-stock lines that have hit their reorder point.
- Load the hours. Units times run time, plus setup per batch. Compare with available capacity at 85%.
- Check the constraint. Separately, in its own units — oven cycles, machine hours, cure days.
- Check materials. Explode the BOMs against stock on hand and confirmed deliveries. Anything short goes on a purchase order today, not on the day it stops the line.
- Sequence. Group by setup similarity, respect due dates, put the constraint's work first.
- Review last week. Planned versus actual hours, and first-pass yield. This is what makes next week's numbers better.
Step 6 is the one that gets dropped and the one that compounds. Standard times drift as processes change, and a plan built on times measured two years ago will be optimistic in a way nobody can see. Comparing planned with actual — even roughly, even occasionally — keeps the whole method honest.