STR.9:4.4 - Find holding regions and meaningful reversals

Challenge a decisive assumption at the lowest effort that can expose its failure. Vary demand, cost, availability, model assumptions or a justified value judgement where those changes can alter the result. Test joint changes when shared causes, thresholds or interactions matter; changing one input at a time can miss them.

Separate the uncertain input, such as future demand, from the response created by a particular service or resource arrangement. If a central-input estimate could hide a material nonlinear response, use PSD.12 section 4.3.1’s paired test or reuse a sufficient qualified analysis. Carry back the response values, tested domain, threshold result and limits on extrapolation. Equal test weights supply no event probabilities, and the same demand can affect two arrangements differently.

State where the option remains acceptable, where a rival becomes preferable and where the comparison is unsupported. Include a condition that defeats all present options when one is material. The least bad candidate is not necessarily admissible.

A new beneficiary, criterion or horizon may define a different comparison rather than a numerical perturbation of the old one. Preserve the earlier basis and make the change explicit. When the option set itself misses a serious response, return to STR.5 or STR.6 instead of treating the current ranking as sufficient.