Segment Typing Tool (Discriminant Classification of New Users)

Description

Uses a few survey questions to assign any new person to an existing persona or segment.

Once segments or personas exist, teams need to recognise members in new studies and recruitment screeners. A typing tool uses discriminant analysis (or logistic regression) on the original segmentation data to find the smallest set of "golden questions" that predicts segment membership accurately. The resulting equations are added to screeners and surveys so every future participant can be tagged with their persona. Sauro and Lewis demonstrate it as the practical payoff of factor and cluster analysis, keeping personas alive and measurable rather than posters on the wall.

Attributes

Loss type
Siloed
Effort
Hours

Steps

  1. Start from a segmentation study with known segment membership for each respondent.
  2. Run stepwise discriminant analysis to find questions that best separate segments.
  3. Check classification accuracy with holdout data or cross-validation (aim for 80% or more).
  4. Reduce to the fewest items that keep accuracy acceptable.
  5. Embed the items and scoring in screeners, surveys and analytics.

Sources

  • Factor and Discriminant Analysis (Sauro & Lewis, MeasuringU training)
  • Practical Statistics for Data Scientists (Bruce & Bruce)

Also known as

  • Persona typing tool
  • Golden questions classifier

Categories

  • Analytics & Behaviour
  • Product Analytics & Strategy