WEEKLY PRODUCTION MIX OPTIMIZER · RUNS IN YOUR BROWSER · NOTHING UPLOADED
What should you make this week?
List what you make, what each product earns and how much of each limited resource (oven hours, labour, machine time) it uses. The optimizer finds the most profitable mix in whole batches, with your promised minimums, your demand limits and optional overtime — and shows what one more hour of your bottleneck is worth and what each product really earns once it crowds out the others.
Arithmetic on figures you supply. Not legal, tax, or financial advice. New to this? Read why your best margin product may not be your best product.
Limited resources (per week)
Anything that runs out: hours of an oven, machine or person, a shared tool, a stock of material. The order matters: product usage columns follow it.
Your products
Unit cost is materials and anything else you pay per unit. Minimum is what you have promised; maximum is the most you can sell (blank for no limit). Batch size is how many come out of one run, tray or case (1 for single units). Use per unit is in the same units as the resource, in the same order as the list on the left.
Add your resources and products, then click Find the best mix.
Result
The plan
| Product | Make | Weekly margin | Margin per unit | Resource cost per unit | Net per unit | Note |
|---|
Resource cost per unit is each resource the product uses times what one more unit of that resource is worth. A negative net means the product earns less than the production it crowds out.
Your limits
| Resource | Available | Used | Use | Overtime | One more unit is worth | Holds for another |
|---|
Against rules of thumb
| Plan | Weekly contribution | Compared with the optimizer |
|---|
What price changes the plan
| Product | Your price | In the plan? | Price at which that flips | Change |
|---|
Each row re-solves the whole problem with only that product's price changed. For a product in the plan it is the lowest price that keeps it there; for one left out, the lowest price that earns it a place. Use it to see how far a price rise or a supplier discount has to go before it changes what you bake.
How it is calculated
Model: maximise the sum of margin (price less unit cost) times units, less overtime cost, subject to each resource's capacity plus overtime, each product's minimum and maximum, and whole batches. The linear program is solved with a two-phase simplex method (Bland's rule, which cannot cycle), and whole batches are found by branch and bound; the page says whether that search finished or stopped at its limit. One more unit is worth: the shadow price, read by re-solving with a little more of that resource, together with how much more it holds for before the plan changes shape; it is a continuous figure for the next unit, not a promise for a whole shift. Resource cost per unit: the resource use times those prices. Rules of thumb: two greedy plans — best margin per unit first, and best margin per unit of your tightest resource first — filled up to your maximums without overtime.
Not modelled: setup or changeover time and cost, minimum run sizes beyond your batch size, ingredients shared between products unless you list them as a resource, shelf life, sequencing within a week, demand that falls as you sell more, taxes and overheads other than the fixed cost you enter, and costs or usage that change with volume. Usage and costs are assumed constant per unit. Unverified against a real business's books: tests use hand-solved problems, brute-force comparison and duality checks.
Common questions
How do I decide what to make when I cannot make everything?
Compare products by what they earn per unit of the thing that runs out, not per unit sold. A product with a large margin that takes most of a scarce oven hour can earn less than a small one that takes a tenth of it. List your resources, the hours or units each product uses, your minimums and your demand limits, and the optimizer finds the combination with the most profit.
What is a shadow price, and why does it matter to a small business?
It is what one more unit of a limited resource would add to profit while the plan stays the same shape: the value of one more oven hour, one more hour of labour, one more machine day. Compare it with what extra capacity costs you (overtime, a second shift, hiring) to see whether buying it pays. The tool also shows how many extra units that value holds for.
Why does the tool charge each product for the resources it uses?
Because a unit margin ignores what the product displaces. The resource cost per unit is the resource use multiplied by the shadow prices; the net per unit is the margin minus that. A product with a negative net is costing you more in crowded-out production than it earns, which is why it is left out of the plan.
Can it handle batch sizes, minimum orders and overtime?
Yes. Give each product a batch size (trays, runs, cases) and it plans in whole batches. Give each a minimum for orders you have promised and a maximum for demand you can really sell. For each resource you can enter overtime hours available and what each costs; overtime is used only where it pays.
Is the answer guaranteed to be the best plan?
For the figures you enter, when the batch search finishes it is the best whole-batch plan, and the page says so. If a very large problem reaches the search limit the page reports how close the plan is to the theoretical upper bound. The result is only as good as your usage figures and costs, which are assumed to be constant per unit.
Is my data uploaded?
No. The optimizer runs in your browser from the figures you type, nothing is sent to a server, there are no accounts, and it works offline after the page loads.