Separate Exploration from Convergence

Learning objective

By the end of this lesson, you can create credible alternatives before evaluating them, then converge using the same criteria and evidence standard.

Why it matters

Many sessions converge twice: first, silently, around the sponsor's favoured answer; later, visibly, around a voting exercise. The apparent exploration only decorates a choice that was already made.

Core concept: do not evaluate while the option space is still forming

Use two distinct modes:

• Explore: widen the option space, surface constraints and create genuinely different responses.
• Converge: test every option against the same mandatory conditions, criteria, evidence and adverse consequences.

The movement resembles the Design Council's divergent and convergent stages, but a working session is not required to produce novelty. A known option, the status quo or a finite diagnostic cycle can be legitimate when stated and assessed fairly.

Define criteria before scoring. Otherwise participants can change what matters after they see which option wins. Separate mandatory conditions from desirable benefits, and review residual risk outside one opaque total.

Visual explanation

Text alternative: the session first creates several real alternatives without scoring, then compares them through mandatory conditions, common criteria, comparable evidence and a risk challenge, and finally closes through a decision, finite evidence gate or explicit escalation.

Worked example: NordForge

NordForge retains four alternatives:

• A — local quick fixes;
• B — a two-site flow pilot;
• C — an enterprise operating system;
• D — a three-month diagnostic cycle.

The group compares customer risk, expected effect, reversibility, change capacity, cost and time to evidence. Before evaluation, each option owner states the strongest case for another option. Only after this steelman round does the group compare trade-offs.

Convergence rules

1. Remove options that fail an agreed mandatory condition.
2. Show evidence beside every remaining judgement.
3. Test the leading option against its strongest credible objection.
4. Record why the runner-up was not selected.
5. Retain a conditional branch if an unresolved fact could reverse the choice.

Do not use dot voting to hide incomparable criteria. A popularity signal can help organise attention, but it cannot establish eligibility, evidence quality or authority.

Common mistake

Mistake: using “do nothing” as a caricature.

The current state has costs and risks, but it is still an option. Define it honestly. A finite diagnostic is different from indefinite delay and should specify what is learned, by whom and by when.

Quick check

The sponsor presents one detailed option and asks the group to “improve it.” What kind of session is this?

A. Open option exploration.
B. Proposal development or consultation, unless the mandate is changed to permit alternatives.
C. Consensus decision making.
D. Evidence review automatically.

Answer: B. The work may be legitimate, but it should not be presented as open choice.

Practical prompt

Write three meaningfully different options for your decision and one explicit status-quo or diagnostic option. Define common criteria before rating any option. Use the Decision Workshop when comparable options already exist.

Summary

• Separate option creation from evaluation.
• Keep at least two credible alternatives visible.
• Define criteria before scoring.
• Apply the same burden of proof to every option.
• Preserve conditional branches when sensitive evidence is unresolved.

Sources

• Design Council, The Double Diamond (opens in a new tab).
• NASA, Systems Engineering Handbook (opens in a new tab), Decision Analysis Process.

Next lesson

Continue with Close with an Explicit Decision Rule: distinguish a decision from a preference poll, vague deferral or hidden escalation.