1) What members share (low burden, optional where possible)

Members can use Healthcount with minimal input: an occasional weight datapoint, an optional medication schedule (dose/day) with lightweight "missed dose" check-ins, and short check-ins when needed. Optional activity/sleep trends can be connected where available.

2) How we turn signals into "maintenance states"

Weight is treated as a lagging indicator, so Healthcount looks for earlier drift signals (for example missed doses, appetite return, side-effect friction, disrupted sleep, reduced activity, reduced confidence). When drift is detected, the system selects one next useful action at a time (or clear signposting to clinical support if that's the right next step).

3) What funders see (aggregate only)

Funders receive grouped, de-identified reporting on:

• Activation (e.g., first check-in within 14 days)
• Retention (e.g., active at 8 weeks)
• Stop–start proxies (e.g., 28+ day gaps and restart rate per 100 member-months)
• Drift signal rates (what's driving friction in the cohort)
• Safety signposting (how often people are prompted to seek appropriate support)
• Burden and trust (opt-outs, opt-out reasons, and typical time spent)

4) Privacy controls (non-negotiable)

• No individual-level reporting to employers
• Aggregated reporting only, using minimum cohort sizes (e.g., 10+ per subgroup) before any metric is shown
• Clear consent and user controls for data connection, export, and deletion
• Governance artefacts produced in parallel (DPIA, UK GDPR privacy/security summary, clinical boundaries)

5) Delivery format (pilot-friendly)

During pilots, reporting can be delivered as:

• A simple funder dashboard (aggregate only), and/or
• Monthly export (CSV/PDF) suitable for governance packs and renewal discussions.