Headline Reach
96%· 973
of sample reachable by at least one intervention
Only 41 respondents fall outside every lever tested
87%· 877
reachable by habit-building alone
The largest single addressable audience
80%· 808
reachable by two or more interventions
Layering, not choosing, is the default
41· 4%
genuinely unreachable by any lever
38 of 41 are Pragmatic Replacers
Intervention Types

Five levers, sized by responsive audience

Intervention × Persona Matrix

Where each intervention lands, by persona

Each cell shows the responsive % for that intervention within that persona. Rank badges (top-right) indicate where the intervention ranks within the persona's five options. Cell shading intensity scales with responsive %, coloured by persona accent.

0% responsive
~80%
100%
Rank badge shows position within persona (1 = primary lever)
Pairwise Overlap

Where two interventions share an audience

The five most strategically significant pairwise overlaps, ranked by combined audience size. These combinations are where a single venture can cover more of the sample than any intervention alone.

Bridge Audiences

Cross-cluster subgroups that unlock wider reach

People who respond to an intervention type outside of their "expected" cluster. These subgroups are why a well-designed intervention can be bigger than its primary persona suggests — and where a secondary cluster can be reached without changing the core design.

Challenge Brief Frames

Three sized frames for hack event scoping

Each frame is anchored on a primary persona pair and an intervention stack. Audience sizes are pulled directly from the matrix above. Design constraints flag the non-obvious rule each brief must respect to avoid missing its target.

Methodology note · how responsiveness is measured

Each intervention type has a composite score built from 6–7 binary conditions drawn from the survey: disposal methods already in use, stated barriers, factor scores (RC1–RC7) from the Varimax-rotated PCA, and attitudinal signals. A respondent scores one point for each condition they satisfy.

Responsive is defined as a score of ≥2 (≥3 for habit-building, where the threshold is higher because the intervention specifically targets dormancy). This is the broad addressable audience. Strongly responsive is the core audience — ≥4 for tech, values and peer; ≥3 for physical; ≥5 for habit-building. Both figures are reported so the reader can see the difference between "potentially movable" and "obvious target".

Additionality is the test that matters most for venture design. For each intervention, we separate respondents who are already doing the target behaviour (selling online, using a collection bank, donating to charity, disposing at high frequency, passing clothes to friends/family) from those who are not. The incremental figure is the genuine new-behaviour audience — what the intervention would actually unlock. Where this number is small relative to the headline responsive count, the intervention is largely a capture play rather than a behaviour-change play.

Thresholds are deliberate and conservative. The responsive threshold is intentionally inclusive so total opportunity is visible; the strongly-responsive threshold narrows to the core. Both matter, and the gap between them — particularly inside the Pragmatic Replacer — carries real design information.