BSF Stage 2 — Frequency Mapping

Description

Assigns decomposed behavioural components to high, mid and low frequency bands and labels each band with concrete behaviours to inform prioritisation.

Frequency mapping is the second stage of the Behavioral Spectral Framework. Components produced by wavelet decomposition are placed on a frequency axis: high-frequency bands represent repeated micro-habits such as checking or scrolling, mid-frequency bands represent routines such as weekly planning, and low-frequency bands represent milestone events such as renewals or annual reporting. Each band is then labelled with the specific features and behaviours that contribute most energy to it. The result is a spectrum showing which rhythms dominate a product and which features serve each rhythm. Its rigour comes from defining band boundaries before analysis, attributing energy to features using event-level data, and comparing spectra across segments rather than relying on a single global picture.

Attributes

Loss type
Community-proposed
Stage
Post-launch
Effort
Days

Steps

  1. Define band boundaries in advance, for example under 1 day, 1 to 14 days and over 14 days.
  2. Aggregate the energy of decomposed components into each band per user and segment.
  3. Attribute band energy to features by recomputing spectra on feature-specific event streams.
  4. Label each band with its dominant behaviours and features.
  5. Compare spectra between retained and churned users or between segments.
  6. Translate dominant and missing rhythms into feature prioritisation hypotheses.

Sources

  • wavelet.pdf

Also known as

  • Behavioural frequency mapping

Categories

  • Community-Proposed Frameworks