Frame the Decision Before Solving It

A decision frame is a concise description of the choice, its purpose, its boundaries, and the people affected. If the frame is wrong, careful analysis can produce a precise answer to the wrong question.

“What should we do about customer support?” is a topic, not a decision. “Which support model should we adopt for UK customers over the next twelve months to cut first-response time below four hours without increasing annual cost above £400,000?” names an action, a population, a horizon, desired results, and a constraint.

Framing is not neutral. “How do we make employees return to the office?” excludes remote and hybrid designs before evaluation. “What working model best supports delivery, learning, and retention?” leaves the means open. The words can quietly decide the answer.

Before researching, write the frame and ask: What answer has this wording already ruled out?

A decision statement names who must choose what and by when. It converts a cloud of concern into a sentence that can guide option generation and evidence collection.

Use this template:

[Decision owner] will decide by [date] whether/how to [action] for [scope] over [time horizon] so that [primary outcomes], subject to [hard constraints].

Compare weak and useful statements:

Weak statement	What is missing	Stronger decision statement
“Should I move?”	Destination, purpose, time	“By 1 October, I will choose whether to remain in London or move to Manchester for at least two years to improve housing stability while preserving weekly family access.”
“Pick a supplier.”	Owner, scope, outcome	“The operations director will select a fulfilment supplier for UK orders by 15 August to support 10,000 monthly shipments with 99.5% dispatch accuracy.”
“Fix onboarding.”	Decision, boundary, criterion	“The product lead will choose the first-week onboarding model for new self-serve customers to raise activation without adding mandatory sales calls.”

Use a verb that represents a choice: select, allocate, approve, stop, launch, hire, postpone, or redesign. Avoid “explore,” “consider,” and “look into” unless the actual decision is whether to fund an investigation.

Run the two-reader test: give the statement to two people separately. If they propose different decisions because they interpret the action or scope differently, rewrite it.

An objective is a result you want more of or less of; a constraint is a boundary an option must not cross. Confusing them can eliminate useful options or make prohibited choices appear negotiable.

Type	Example	Treatment
Objective	Reduce response time	Prefer more improvement, all else equal
Objective	Improve employee learning	Compare degree of benefit
Hard constraint	Must comply with UK GDPR	Disqualify any violating option
Hard constraint	Cannot exceed £400,000 approved spend	Disqualify or redesign
Soft constraint	Prefer launch before Christmas	Treat as a weighted objective unless truly non-negotiable
Assumed constraint	“Customers will never use chat”	Test with evidence before accepting

The word “must” deserves suspicion. Ask who imposed the boundary, what happens if it is crossed, and whether a waiver or redesign exists. A legal requirement and an executive preference should not occupy the same category.

Also watch for objectives disguised as solutions. “We need an AI chatbot” is a proposed means. The objective may be “answer routine questions accurately at any hour.” Once restated, better options can enter.

A stakeholder is a person or group whose outcomes, knowledge, or responsibilities are materially affected by the decision. Mapping them prevents the owner from optimizing only the most visible result.

For each stakeholder, record three things:

Stakeholder question	Purpose
What outcome do they need protected?	Reveals values missing from the frame
What information do they possess?	Locates evidence and practical constraints
What consequence do they bear?	Exposes costs transferred away from the owner

In the support-model example, leaders may value cost and retention, customers value fast accurate resolution, support agents value workload and tools, security staff carry data risk, and finance controls budget. None automatically receives a veto. Their interests enter the frame; decision authority remains explicit.

Distinguish consultation from consent. Consultation means the owner must hear relevant knowledge and effects. Consent means a party has authority to block. If that distinction is vague, the process may either ignore people or stall indefinitely.

Scope states where the decision applies, while the time horizon states how long its consequences should be judged. Both prevent an option from looking good only because important effects were placed outside the picture.

A narrow horizon can make a cheap system attractive while ignoring migration costs next year. An overly broad horizon can paralyze a reversible three-week experiment with speculation about ten years.

Use four boundary questions:

1. Population: Which customers, teams, products, or locations are included?
2. Decision period: When must the choice be made?
3. Evaluation horizon: Over what period will costs and benefits count?
4. Adjacent decisions: Which related choices are explicitly outside scope?

Boundary	Support example
Population	UK direct customers; enterprise accounts excluded
Decision period	Select model by 15 August
Evaluation horizon	Compare total effects over twelve months
Outside scope	Replacing the company-wide CRM

Write exclusions down. “Enterprise support is outside this decision because a separate contractual review owns it” is safer than silently ignoring it. A named exclusion can be challenged; an invisible one cannot.

A criterion is a standard used to compare options. Strong criteria connect a value to observable evidence so different people can understand why one option scored above another.

Move through three levels:

Value	Decision objective	Observable criterion
Customer care	Resolve problems quickly	Median and 90th-percentile resolution time
Reliability	Give correct answers	Audited answer accuracy and reopen rate
Sustainability	Avoid exhausting staff	Cases per agent, overtime hours, absence rate
Stewardship	Use money responsibly	Twelve-month total cost, including transition

Avoid double-counting. “Fast response,” “short wait,” and “quick resolution” may reward the same benefit three times. Define each criterion’s distinct role.

Specify direction and measurement:

• “Resolution time: lower is better; target median below eight hours.”
• “Accuracy: higher is better; floor of 95% on audited cases.”
• “Cost: lower is better; include licences, staffing, training, and migration.”
• “Employee effect: avoid more than two overtime hours per agent per week.”

Not every value can be reduced to one number. Qualitative scales are legitimate if anchors are clear: a score of 5 for maintainability might mean two trained internal teams can operate the system with documented procedures; a score of 1 might mean dependence on one external specialist.

An assumption is a claim the frame treats as true without yet proving it. Writing assumptions early prevents research from becoming a pile of facts that never challenge the logic of the choice.

Create an assumption register:

Assumption	If false, what changes?	Current confidence	Cheapest useful test
Half of requests are routine	Automation option loses value	Medium	Classify 300 recent tickets
A supplier can migrate in six weeks	Launch date may fail	Low	Request a dated migration plan
Customers accept chat for simple issues	Channel mix changes	Low	Run a two-week opt-in pilot
Two internal teams can operate the tool	Vendor dependence rises	Medium	Ask teams to complete a setup exercise

Prioritize assumptions by decision impact × uncertainty. A highly uncertain fact that would not change the choice is interesting but not urgent. A moderately uncertain fact that could reverse the choice deserves attention.

Frame research as a decision test: “If the routine-request share is below 20%, automation is removed from the shortlist.” This tells you when evidence is sufficient and protects against endless analysis.

Reframing means writing the decision from another legitimate perspective to reveal omitted options or values. It is a diagnostic technique, not wordplay.

Try four reframes:

Original frame	Deliberate reframe	What it may reveal
“Which supplier should we buy from?”	“Which capabilities should we own, rent, or combine?”	Build-and-buy hybrids
“How do we reduce support cost?”	“How do we prevent avoidable support demand?”	Product and documentation fixes
“Should we launch in September?”	“What evidence must be true before launch?”	Readiness gates
“Which applicant is best?”	“Which team design best covers the work?”	Role redesign or internal movement

Do not keep all frames. Compare them and choose the one that best captures the real authority, outcomes, boundaries, and timing. Record one sentence explaining why: “We framed the problem around preventing demand because ticket analysis shows one product defect creates 28% of contacts.”

A worked frame shows how the pieces constrain and improve one another. Imagine a small retailer choosing how to handle fulfilment as orders grow.

Initial topic: “We need a new warehouse.” This already assumes ownership or leasing is the solution. The team reframes it as:

The operations director will decide by 30 September how to fulfil UK orders for the next two years so that the company can ship up to 10,000 orders per month with at least 99.5% dispatch accuracy, while keeping two-year total cost below £1.2 million and remaining compliant with product-handling rules.

Stakeholders: customers bear late or incorrect delivery; warehouse staff bear workload and safety effects; finance bears cash exposure; product teams need launch capacity; the operations director owns the choice.

Options now visible: expand the current site, lease a larger site, outsource to a third-party logistics provider, or combine internal fulfilment for complex products with outsourced standard orders.

Criteria: peak capacity, accuracy, two-year total cost, implementation time, flexibility, staff impact, and compliance control. Assumptions: the forecast peak is credible, providers can handle unusual products, and migration can complete before the holiday period.

The frame does not select the option. It creates a fair field on which relevant options can be compared.

A frame-quality audit is a short check performed before expensive research or debate. It catches hidden solutions, false constraints, missing people, and horizons that distort consequences.

Complete this one-page canvas:

Element	Your answer
Decision owner	Who has authority and accountability?
Decision statement	What action will be chosen, by when, for whom?
Primary outcomes	Which results matter, in ranked order?
Hard constraints	Which boundaries truly disqualify an option?
Stakeholders	Who supplies knowledge, experiences effects, funds, or carries risk?
Scope	Which population, geography, product, and adjacent choices are included?
Time horizon	When is the choice due, and how long do effects count?
Criteria	What observable standards compare options?
Critical assumptions	Which uncertain claims could reverse the decision?
Rationale	Why is this the most useful frame?

Final checklist:

• [ ] The statement describes a decision, not a topic or preferred solution.
• [ ] One owner and one decision date are named.
• [ ] Objectives are distinguished from hard and soft constraints.
• [ ] Stakeholder effects include people who do not control the meeting.
• [ ] Scope exclusions are explicit and owned elsewhere.
• [ ] The evaluation horizon is long enough to capture delayed costs.
• [ ] Criteria are distinct, observable, and directional.
• [ ] Critical assumptions have tests and decision consequences.
• [ ] At least two alternative frames were considered.
• [ ] The selected wording does not quietly rule out the best class of option.

A good frame does not guarantee a good choice. It makes the real choice visible, keeps analysis relevant, and gives disagreement a precise place to land.