Decision, Disagreement & Consensus Board

SAMPLE method: a 10-persona qualitative depth panel derived from the target audience and frozen as a quota matrix before the run. Options are shown blind, symmetrically and in randomized order. Personas answer independently, rejection reasons come first, comprehension is a hard gate (>2/10 material misreads disqualifies an option), aggregation uses median/range plus best-worst and segment/polarization cuts, and a diversity audit checks for mode collapse. Directional only, not statistical.

Synthetic panel. Directional read of objections, not a forecast of demand.

• Synthetic personas can share model biases.
• No real buyer behavior was observed.

• Small qualitative synthetic panel (SAMPLE n=10).
• Variant C wording differed in length.

Comprehension is checked first as a disqualifying gate. Eligible options are summarized with median + range, a forced best-worst tally, segment cuts and polarization; disagreement is reported rather than averaged away.

Choosing which offer to put in front of real traffic first.

Directional median objection strength by variant (SAMPLE, 0–3).

Score scale: 0 none · 1 weak · 2 moderate · 3 strong

[
 [
  "Price feels high",
  1,
  3,
  1
 ],
 [
  "Unclear deliverable",
  2,
  1,
  2
 ],
 [
  "Trust in reviewer",
  2,
  2,
  1
 ],
 [
  "Subscription fatigue",
  0,
  0,
  3
 ]
]

• “Review” read as code review by some personas.
• “48h” unclear: start or delivery?

• “10 prioritized design fixes in 48 hours”
• “One-off, no subscription”