Purpose

This page records selected engagements involving failed or unstable conjoint analysis where intervention focused on containment, decision triage, or prevention of irreversible decision error.

Conjoint rescue is not an extension of conjoint execution. It is an independent intervention applied when standard best-practice methods increase decision risk rather than reduce it.

For the framework describing when and why conjoint rescue is appropriate, see Conjoint Rescue.

Selected Engagements (Record)

Entries are intentionally concise and exclude outcomes. Rescue work often centers on identifying what cannot be safely inferred, rather than improving model performance.

Engagement 1 — Irrecoverable Experimental Design Failure

• Context
  Technology platform planning a future major version, with widely varying development costs and unknown consumer value across a large innovation set.
• Failure Mode
  Late-stage utilities implied effectively unbounded consumer value across features, with price sensitivity collapsed in a way that could not be resolved through re-estimation, tuning, or vendor-supported diagnostics.
• Nature of Rescue
  Design-level triage revealed a structural experimental flaw—assumed to exist elsewhere but actually embedded in the design—that made price sensitivity unrecoverable for a subset of respondents. Rescue required model restructuring and explicit exclusion of contaminated data rather than rerunning the study.

Engagement 2 — Evidence-Frame Deadlock Under Vendor Escalation

• Context
  Healthcare software company pursuing portfolio pricing and bundling decisions using standard conjoint tooling and accepted best practices.
• Failure Mode
  Despite reruns and escalation, analysis remained confined to feature importance outputs that could not answer pricing or bundling decisions. Importance was treated as a proxy for pricing strategy, and guidance remained unchanged because failure was interpreted as incorrect use rather than a mis-specified decision frame.
• Nature of Rescue
  Governance triage surfaced an evidence-frame deadlock: even perfectly estimated importance measures could only serve as contextual input, never as decision-grade pricing evidence. Rescue required formally demoting importance to context and restructuring the analysis around explicit portfolio trade-offs before executive exposure locked in the wrong evidentiary standard.

Notes

These rescues did not arise from choosing the wrong method; conjoint analysis was the natural and appropriate starting point for the decisions involved. Many rescue engagements originate from studies built by capable vendors using accepted methods, where failure emerges only when outputs meet real decision, portfolio, or governance constraints.

Because rescue work often requires containing or demoting previously delivered outputs—rather than refining them—it is rarely performed by the same teams that designed or executed the original study.

Why this page exists

This record exists to document experience, not outcomes.