Planning
Calculated, not argued.
Three questions, one model: can the operation deliver the plan, which investment is worth it, and exactly what should be set?
Most capital decisions are argued because calculating them used to cost more. This gives judgement something to work on.Most capital decisions in manufacturing are argued rather than calculated, because calculating them properly used to cost more than the argument did. That is the gap this closes: not by replacing judgement, but by giving it something to be exercised on.
Will the operation deliver the plan?
Run the volume you promise and see what breaks first: often not the constraint everyone watches. A model shows it in a week, not a year.A growth target is a claim about a physical system. Run it at the volume you are promising and see what breaks first, and how far before it does. Often the constraint that binds first is not the one everyone watches, and finding that out in a model costs a week, not a year.
- The constraint that actually caps throughput, not the loudest one
- An ambitious target told apart from an arithmetically impossible one
- What a change in demand or product mix does, before you commit to it
Every option on one set of assumptions
Run every option through one model and you compare options, not business cases. Any difference chance could explain is marked as noise.Investment arguments go in circles because each option arrives with its own business case, built by whoever is advocating it. Run them all through the same model and the comparison is about the options, not the spreadsheets. Every difference that random variation could explain is marked as noise rather than argued over.
- Up to five options against one reference you pick
- Each scored again against that reference, so options from different studies can be weighed together
- Six sections: economic, operational, quality, delivery, material and sustainability
Where the lead changes hands
Weight cost and delivery your way: the ranking responds at once, with no new runs, up to the exact point where the lead changes hands.Two directors weighting cost and delivery differently will reach different conclusions from the same data, and that disagreement is legitimate. Move the weighting and the ranking responds at once (nothing is simulated again; only the weights change), up to the exact point at which the lead changes hands.
- Four priority profiles ship (balanced, or led by cost, by throughput or by sustainability), and you can add your own
- How far each priority has to move before the leader changes, or that it leads at every weight
- A breakeven is measured like any other figure: more runs can move it, and you can check
What it earns back
Declare cost, life, end value and servicing, and the same run appraises the machine; servicing is a planned stop traded against breakdowns.Declare what a machine costs, how long it lasts, what it is worth at the end and how it is serviced, and the same run gives the appraisal, with no separate spreadsheet. Servicing is a real planned stop, so servicing more often trades stops against breakdowns instead of costing nothing.
- Between two options: the extra capital, the extra cash a year, NPV, IRR and payback
- Each one a measure like any other: in the dashboard, in reports, and something a goal can be set on
Say what “better” means first
The goal decides what a search finds: pick a named one or write your own. Most bring a guardrail holding what they would otherwise give up.A search finds exactly what it is asked for, so the goal decides the answer. Pick a named goal (lowest cost per good unit, most output, best delivery on time, fewest emissions) or write your own as a plain expression. Most goals bring a guardrail that holds what they would otherwise give up, measured against today.
- Set a limit of your own and the guardrails step aside: nothing applies that you cannot see
Let it search, and do not trust the lucky winner
Declare levers and limits and let it search. The shortlist is run again properly, so the figure reported is never a lucky screening result.Declare the levers and the limits, and the software searches: a surrogate search that fits a surface through every point it has seen, random search, simulated annealing or the full grid. Picking the best of many noisy estimates picks partly for luck, so the shortlist is run again properly, and the number reported comes from there, never from screening.
- 01
Declare levers and limits
How many operators, how big a buffer, which shift pattern, and what the answer must respect. A limit on a setting alone is checked before anything runs.
- 02
Screen wide and cheap
Many points, a few runs each, ranked on their average: enough to find the contenders, not to crown one.
- 03
Decide the shortlist properly
The best few run again with full replications. Each limit is judged at the unfavourable end of its interval, so a point that keeps it only on average does not pass.
- 04
Report what mattered
The winner, whether it can be told apart from the second best, and which levers moved the result at all.
The same model, next
Once the configuration is set, the same model goes on to plan the days.
Simulation
Build a working model of your operation and run it as the real thing.
Operations
Shifts, rosters, breaks and calendars, and the daily decisions they drive.
Analysis
Product passport, weighing the options, proposed actions and reports both audiences can read.
Energy
Every hour priced on its own, the grid's CO₂ by the hour, and your own solar and wind.
Sustainability
Emissions, energy, water and material, answered in the same run as cost.
What decision is on your desk right now?
Bring the one you are least sure about. We will walk through what the model would show you, and say plainly if it would not help.
Book a discovery call(opens in a new tab)