A decision skills case study: rewriting one invented order choice until it can be decided

This is one decision, filled in field by field. It is invented, and no figures appear, because the useful part is watching the framing change the answer twice before any evidence arrives.

The situation: a small workshop that makes and installs a physical product is asked by a long-standing customer to take on a much larger recurring order, and would have to add a second shift to accept it.

What to take away

The first version of a decision is almost never the decision. Two rewrites here change what is being chosen, and neither rewrite required new information.
The criteria are where the disagreement lives. Writing them before the options are compared is the single highest-return habit in the whole exercise.
Record what would change the answer. That sentence is what makes the review possible a year later.

The decision as first stated

"Should we take the big order?"

This is a topic wearing a question mark. It has one option in it, no decider, no date, and no statement of what taking it would require. The twelve reframed decision statements cover this shape: a yes or no where the interesting question is what problem is being solved.

Rewrite one

"Whether to accept the order at the offered terms, decided by the owner, by the end of the month."

Better. It now has a decider and a date. It still contains a false binary, because declining is not the only alternative to accepting as offered.

Rewrite two

*"Which of these to do by the end of the month: accept as offered, propose

Three versions of the decision

First stated

Options: One
Decider: None
Date: None
Decidable: No

Rewrite one

Options: One
Decider: Owner
Date: End of month
Decidable: No

Rewrite two

Options: Four
Decider: Owner
Date: End of month
Decidable: Yes

Four options, and the third and fourth are genuinely different from each other. This is the version that can be decided. Notice that nothing was learned between the first version and this one. The work was entirely in the statement.

The objective, before the options are compared

Write this before comparing anything, or the objective will quietly become whatever the attractive option is good at.

For this workshop the objective might be: to increase what the business earns over the next two years without putting the existing customers at risk. That sentence already rules things out, which is what an objective is for.

The constraints

Constraints that eliminate options

Cash before first payment arrives
Second shift staffable locally
Commitments to existing customers
Constraint eliminates, criterion ranks

about profit.

Constraints are not criteria. A constraint eliminates an option; a criterion ranks the survivors. Mixing them is how a decision ends up with a preferred option that cannot actually be done.

The criteria, and the argument they contain

Criterion	Why it might matter	Who would weight it highly
Total earnings over two years	The stated objective	The owner
Cash timing	Whether you survive the ramp	Anyone who has run out of cash
Customer concentration	Dependence on one buyer	Anyone who has lost a large customer
Quality risk during ramp	Reputation with existing buyers	The people doing the work

The third and fourth rows are where a real disagreement would sit, and it is a

disagreement about risk tolerance rather than about facts. Naming it as such is what stops it being argued as though someone were wrong about arithmetic.

The evidence, and its limits

Some of what is needed is knowable: the terms offered, the current cost

Knowable versus unknowable evidence

Knowable now

Terms: Offered terms
Costs: Current cost structure
Wages: Local wage rates
Customers: Existing commitments

Not knowable now

Terms: Future volume demand
Costs: Ramp quality
Wages: Second-shift staffing
Customers: Deprioritized buyers' return

Write both lists. The second list is the one that matters, because it is where the decision is actually being made, and pretending it belongs to the first list is how a confident answer gets produced from nothing. The distinction between what a figure records and what it assumes is set out in the metric words guide.

Reversibility

Accepting is partly reversible and partly not. The order can be ended on its

terms; the hires cannot be unmade without cost, and the existing customers who were deprioritized will not necessarily come back.

That asymmetry is the finding. It suggests a fifth option nobody listed: accept a portion, on a fixed trial, with the terms for the rest agreed in advance. The roles and process guide calls this converting a hard decision into two easy ones, and it is available far more often than it is used.

What would change the answer

Written at the time, before the outcome exists:

If the customer will not agree to a minimum volume, decline. If a second shift cannot be staffed by a named date, decline. If both hold, accept the trial portion.

Three sentences. A year later, this is the only thing that makes it possible to tell a sound decision from a lucky one, which is exactly what the decision review measures depend on.

Common questions

Is this much process worth it for one order?

For a decision that involves hiring and a large customer, yes, and it is about thirty minutes of writing. For a small reversible choice, no. The sizing question is reversibility, not importance.

What if the owner has already decided?

Then the exercise is still worth running once, because it usually surfaces the fifth option. If it changes nothing, that is a legitimate outcome and it took half an hour.

Where do the outside figures come from?

From the customer, from the accounts, and from public sources for anything general. Guidance on approaching that research, such as the Small Business Administration's market research pages, is a reasonable starting point. Nothing on this page supplies a figure, and nothing should.

Why does the framing change the answer so much?

Because a badly stated decision hides options, and hidden options cannot be chosen. Adding the fourth and fifth here cost nothing and produced the answer most people would want. The nine decision errors list the framings that hide options most reliably.