Professional / Lark Health
Enrollment forecasting at Lark Health
I built one-year enrollment forecasts for Carelon, BCBSAL, and MMO using audience size, marketing sends, and seasonality. The team uses them to plan send volumes and direct-mail budgets, with email costs also considered.
- Role / scope
- Lead Data Analyst, Marketing. Built the partner forecasts now used in planning.
- Timeframe
- In use during 2026; results as of September 2026
- Business problem
- How many enrollments should each partner’s outreach generate, and what does that imply for send volumes and channel budgets?
- Key result
- Monthly enrollment error, checked separately for each partner and accounting for backdated enrollments, has usually been within 10% in 2026 to date, as of September. The forecasts are used in outreach and budget planning.
Forecast inputs and seasonality
I built the forecasts from data pulls showing the relationship between audience or list size, marketing sends, and resulting enrollments. I accounted for seasonality and projected one year ahead for Carelon, BCBSAL, and MMO.
Inputs
- Audience and list sizes
- Marketing sends
- Observed enrollments
Forecast
- Partner-specific relationships
- Seasonality
- One-year horizon
Decisions
- Send volumes
- Direct-mail budgets
- Email costs considered
Monthly checks for each partner
I compare forecast enrollments with actuals each month, separately for Carelon, BCBSAL, and MMO, after accounting for backdated enrollments.
As of September 2026, monthly enrollment error has usually been within 10% for each partner during the year. The underlying monthly series and a baseline comparison are not published here.
Outreach and budget planning
The forecasts inform how much outreach to send and how to budget direct mail. Email costs are also considered in planning, even though email is often treated as free.
These forecasts project enrollment from observed relationships and do not estimate causal lift. The team also uses the attribution model to review where past enrollments came from.