Assumption Stress-test Workshop¶
Facilitated workshop — instantiates Sensitivity Analysis Protocol
Convenes the people who own or dispute the assumptions to argue defensible ranges, name the decision-carrying ones, and set the validation agenda.
The assumption stress-test workshop is the human, facilitated mechanism of the protocol. Rather than computing anything, it convenes the people who own or contest the assumptions and structures a session to do three things a solver cannot: surface which assumptions actually carry the decision, set defensible ranges by argued judgment rather than analyst convenience, and name what must be validated before committing. Its defining move is social: it treats range-setting and criticality as contested claims to be adjudicated by domain experts and stakeholders. That is the only reliable way to handle assumptions that are value-laden, political, or simply unmeasured — the ones no distribution or grid can settle because the disagreement is about judgment, not arithmetic.
Example¶
Two companies are negotiating a merger, and the board vote hinges on a valuation model whose bottom line is dominated by a few soft assumptions: how much cost synergy the combined firm will actually realize, how long integration will take, and how many key customers will defect during the transition. The deal team's model shows an attractive number, but the assumptions behind it were set by the people who want the deal. So the CFO runs an assumption stress-test workshop. Around the table: the deal team, operators from both firms, an integration lead, and a skeptic from finance. For each contested assumption the group argues a defensible range — synergy realization is pushed down from the deal team's 90% to a debated 55%–80% band — flags which assumptions would sink the deal if they landed at the low end, and leaves with a validation queue: commission a customer-retention survey, get an independent integration-cost estimate, and revisit before the vote. Nothing was computed; the inputs to every downstream calculation were made honest.
How it works¶
- Assemble owners and skeptics. Bring the people who hold, dispute, or will be judged on the assumptions into one room, with a facilitator and a deliberate dissenting voice.
- Nominate the carrying assumptions. Elicit and debate which assumptions actually decide the outcome, rather than accepting the model's given list.
- Argue defensible ranges. For each, negotiate a low-to-high band the group can defend, separating independent elicitation from open discussion to blunt anchoring.
- Leave with a queue. Produce a prioritized list of assumptions to validate or monitor, with an owner and a deadline for each.
Tuning parameters¶
- Participant mix — broad and adversarial versus narrow and aligned; breadth surfaces more blind spots but is harder to converge, alignment converges fast but can rubber-stamp.
- Elicitation structure — open discussion versus Delphi-style independent estimates reconciled afterward; independence fights anchoring but costs time.
- Range-setting basis — data-anchored where evidence exists versus pure expert judgment where it does not; the mix sets how defensible the resulting bands are.
- Adversarial framing — neutral review versus an explicit premortem that asks "assume this failed — which assumption was wrong?"; the harder framing surfaces more but can feel accusatory.
When it helps, and when it misleads¶
Its strength is that it handles exactly the assumptions no computation can: contested, value-laden, or unmeasured inputs where the honest range is a matter of judgment, and where stakeholder buy-in to the ranges is itself worth having. Its failure mode is groupthink and anchoring — the loudest or most senior voice sets the band, or the room converges on a comfortable consensus, so range-setting is laundered by committee.[n1] The classic misuse is a workshop convened to ratify the sponsor's preferred assumptions under the appearance of scrutiny. The guarding discipline is independent elicitation before any discussion, an assigned devil's advocate, and recording dissent rather than smoothing it away, so the ranges reflect argued judgment instead of social pressure.
How it implements the components¶
key_parameter— the group nominates and debates which assumptions genuinely carry the decision, rather than inheriting a fixed list.variation_range— sets each assumption's low-to-high band by argued, defensible expert judgment in the room.critical_assumption— records the group's verdict on which assumptions would break the decision if they moved to the wrong end.validation_follow_up_queue— its deliverable: a prioritized, owned list of assumptions to validate or monitor before committing.
It sets inputs but runs none of the numbers: it does not sweep parameters to rank them (parameter_prioritization_rule, One-way Sensitivity Analysis), map interaction surfaces (correlation_or_dependency_note, Two-way or Multi-way Sensitivity Analysis), or solve for a break-even value (decision_flip_point, Threshold Analysis). The quantitative mechanisms consume the ranges it produces.
Related¶
- Instantiates: Sensitivity Analysis Protocol — this is the protocol's human range-setting and prioritization mechanism, upstream of the computational ones.
- Sibling mechanisms: One-way Sensitivity Analysis · Two-way or Multi-way Sensitivity Analysis · Scenario Variation · Threshold Analysis · Tornado Chart · Sensitivity Table · Probabilistic Sensitivity Simulation
Editorial Notes¶
Form Classification¶
Form family: Communication, Facilitation & Learning
Rationale: Convenes the people who own or dispute the assumptions to argue defensible ranges, name the decision-carrying ones, and set the validation agenda, making its operative form a designed exchange or learning exposure that changes understanding or capability.
Independent corroboration: The frozen evidence defines Assumption Stress-test Workshop as 'Convenes the people who own or dispute the assumptions to argue defensible ranges, name the decision-carrying ones, and set the validation agenda', so its operative form is Communication, Facilitation & Learning.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Statistics & Experimental Design
Origin pattern: Cross-disciplinary synthesis
Present-day reach: Multi-domain
Rationale: Structured expert elicitation and sensitivity analysis provide the defining operations: elicit defensible ranges, expose uncertainty, and compare consequences.
Related originating lineages:
- Futurism & Strategic Foresight — Scenario variation supplies plausible future conditions against which assumptions are challenged.
- Operations Research — Sensitivity analysis and decision modeling structure the consequence comparison.
- Organizational & Management Science — Workshop facilitation, ownership, and commitment convert elicited ranges into coordinated action.
Review resolution: Authoritative guidance on structured expert elicitation treats expert judgments as data gathered through a protocol, while the IDEA protocol formalizes independent estimates, discussion, and revision. Those operations define the workshop more strongly than its facilitation wrapper, so statistics and experimental design are primary.
Attribution caveat: The statistical elicitation core is wrapped in an Encyclopedia-authored management workshop that also draws on scenarios and operations research.
Encyclopedia synthesis: The exact catalogued form synthesizes established practice rather than reproducing a single standard historical label.
Review outcome: Researched adjudication after independent review; high confidence.
Sources consulted:
- NCBI Bookshelf — Structured Expert Elicitation
- Methods in Ecology and Evolution — The IDEA Protocol for Structured Expert Judgment
Notes¶
[n1] The Delphi method structures expert input by collecting estimates independently and then feeding back anonymized summaries over rounds, deliberately preventing the face-to-face dynamics — deference to seniority, anchoring on the first number spoken — that distort group judgment. It is the standard corrective when a workshop's ranges must reflect genuine expert judgment rather than the loudest voice. ↩