01 · What a must criterion does
What a must criterion does
Must criteria are conditions an option has to meet to remain eligible. They create a feasibility gate before trade-offs between desirable features. A solution that saves time but blocks an emergency route cannot compensate for that failure by scoring highly on cost. A solution that breaks required shipment traceability is not rescued by enthusiastic voting. The purpose is to separate genuine constraints from preferences that can legitimately be balanced.
Use must criteria before a decision matrix, Pugh comparison, effort–benefit discussion, or pilot approval. Prerequisites include a defined decision, people who understand the relevant requirements, and a way to verify each condition. “Easy to use” is usually a preference until the team states a testable minimum. “A trained operator can complete the recovery sequence without administrative access” is a possible gate if that capability is essential for continuous dispatch.
02 · Build requirements with evidence
Build requirements with evidence
Fictional Northbank Dispatch considers four concepts: fixed carrier lanes, early lane assignment with scan confirmation, a new conveyor, and a large shared information screen. Its sponsors want fewer late collections, but safety, traceability, installation disruption, and the authorized spending ceiling also constrain the choice. The team asks the safety representative, warehouse owner, and systems owner to identify the actual requirements and the evidence supporting them.
It writes five local gates: preserve the approved aisle clearances; retain carton and carrier identity in the dispatch record; provide an approved manual fallback; fit a £15,000 initial spending authorization; and complete installation within one planned weekend closure. The budget and closure are local project constraints, not universal lean requirements. Each gate names the verifier. This is essential because a meeting participant saying “probably fine” is different from the responsible owner accepting documented evidence.
03 · Screen with pass, fail, and unresolved
Screen with pass, fail, and unresolved
For every option, assess each gate as pass, fail, or unresolved. Unresolved is not a disguised pass. Record what evidence is missing and whether the option can be modified to satisfy the requirement. A fixed-lane design might pass the cost limit but need a safety layout review. A scan-confirmation design might preserve traceability in normal operation but still lack an offline recovery procedure.
The team keeps desirability scoring out of this discussion. Doing otherwise encourages people to overlook a failed condition because they like the option’s benefits. It also checks whether a supposed must is truly essential. “Must use the current screen supplier” may merely be a preference, while “must integrate with the existing dispatch identifier” may be indispensable. Reclassifying an unjustified gate increases the choice set without relaxing the real constraint. Changes to gates are documented before the final comparison, not after an inconvenient option fails.
04 · Worked screening example
Worked screening example
The conveyor quote is £24,000 and needs three closure days, so it fails two current gates. It is parked for a separate capital decision rather than described as a poor technical idea. The information screen costs £6,000 and fits the closure, but by itself does not preserve positive identity checking; operators can still place a carton in the wrong lane. It remains eligible only as an information component, not as a complete identity solution.
Early assignment with scan confirmation is quoted at £12,000 and requires one weekend. Its manual fallback is unresolved. The systems owner demonstrates a controlled paper manifest with carton identifiers and reconciles a sample before the option receives a pass. Fixed lanes cost £4,000 and pass after a layout review. The team now compares two complete eligible concepts. It has not proved either will improve dispatch; it has established that testing them is consistent with the stated constraints.
05 · Interpret the gate and preserve accountability
Interpret the gate and preserve accountability
A pass means the evidence meets the requirement under stated conditions. It should say which version, operating range, and assumptions were reviewed. Northbank’s fallback pass applies to one approved manifest procedure and trained staff; it does not automatically cover every handwritten workaround. A fail means the current design is ineligible under the current decision authority. An unresolved assessment means the team needs evidence or a design change before proceeding.
Owners retain responsibility for their gates during pilot and implementation. The layout verifier reviews any change in lane dimensions; the systems owner reviews identifier changes; the sponsor controls the spending ceiling. If an option evolves after selection, screen the revised version again. This prevents a feasible proposal from becoming an infeasible implementation through small unreviewed changes. Gates also identify pilot guardrails: traceability and aisle access must remain acceptable while benefit is being learned.
06 · Pitfalls and advanced adaptation
Pitfalls and advanced adaptation
Weak practice labels every management preference a must, making innovation impossible. Another weak approach uses broad statements such as “safe” without naming the relevant requirement or verifier. Strong practice uses a small, justified set of gates, distinguishes uncertainty from failure, and gives rejected options a clear explanation. It avoids numerical averaging: four passes cannot cancel one failure on a non-negotiable requirement.
Advanced teams use conditional eligibility. A concept may proceed only for a particular carrier, shift, or carton size while evidence is gathered for other conditions. These boundaries are part of the decision, not footnotes. They also distinguish hard external constraints from limits the sponsor can legitimately revisit. A spending authorization may be changed through a new business case; a required safety clearance cannot be traded away by the project team. This distinction supports practical decision making while preventing the gate from becoming either arbitrary bureaucracy or a rubber stamp.
07 · Operating conditions and evidence
Operating conditions and evidence
A gate can be conditional without being weak. Northbank’s manual fallback is accepted only when the trained shift lead is present and the approved manifest is available. If either condition is absent, the process stays on the current approved method. Recording those conditions is stronger than a blanket statement that the option “has a fallback.” The same logic applies to carton sizes, carrier types, and access permissions.
The screening record also distinguishes evidence expiration from design failure. A layout pass may need review after a new rack is installed, and a systems pass may need review after an identifier change. The team states these review triggers at selection. During implementation, the project owner checks whether the option being built is still the option that passed. This prevents an apparently minor substitution, such as a different scanner model or a revised label format, from bypassing the requirement assessment that made the original concept eligible.
08 · Decision record and next step
Decision record and next step
Northbank retains the must-criteria record with the selected design. This is more useful than filing only a final pass mark. The record explains the specific requirement, the evidence reviewed, the verifier, and the conditions under which the pass remains valid. A new project owner can then recognize when a design change requires renewed review. Without that context, a historical pass can be misused as permanent authorization for a different operating method.
The sponsor also records what would need to change for a parked concept to return. The conveyor can be reconsidered under a larger capital decision and a different closure plan; it is not permanently excluded from all future improvement. The traceability requirement remains unchanged. This distinction makes screening constructive: options receive a route for resolving legitimate constraints, while non-negotiable requirements remain protected. The team can move quickly with eligible options because it has already made the important constraints and evidence explicit.
| Gate outcome | Meaning | Next action |
|---|---|---|
| Pass | Evidence meets the stated condition for this design | Proceed to comparison while retaining review triggers |
| Fail | Current design violates a required condition | Redesign or seek a legitimate change in decision scope |
| Unresolved | Necessary evidence is missing | Obtain verification before treating the option as eligible |
Connections
Related methods and phase use
Placed under Improve because of the outcome it supports. Use the method in other phases when its purpose and assumptions fit the work; the phase list is a guide to selection, not a compulsory sequence.
Sources
Professional foundations and further reading
The explanation and fictional teaching examples are written for LeanTrainingLab. Linked primary sources provide professional foundations; example values and local conventions are stated in the guide. The accepted toolset draws on Six Sigma+Lean Toolset, second edition (Lunau, editor; Springer, 2013). No book text or examples are reproduced here.