Experience

I am the sole owner of marketing analytics at Lark Health. My work covers attribution, enrollment forecasting, lifecycle measurement, paid media, direct mail, activation, and partner programs.

Attribution and channel measurement

I rebuilt production acquisition attribution after the existing system had accumulated independent first-match classifiers, hard-coded rules, and incomplete tracking. Around March 2026, on a trailing 365-day view, about 42% of enrollments were unresolved. The rebuilt model reduced unknown attribution to about 8%.

For one partner population in Q4 2025, direct mail’s measured share of enrollments moved from about 4% under the previous measurement to about 38% after corrected attribution. My findings reversed a decision to discontinue direct mail, a channel with $550K in annual spend. This is a measurement correction, not an incrementality result. Read the attribution case.

Enrollment forecasting and channel planning

I built and shipped one-year enrollment forecasts for Carelon, BCBSAL, and MMO using the observed relationship between audience size, marketing sends, and enrollments, with seasonality accounted for. The forecasts guide send volumes and direct-mail budgets, with email costs also considered in planning.

I check monthly enrollment error separately for each partner and account for backdated enrollments. Error has usually been within 10% during 2026 to date, as of September 2026. Read the forecasting case.

Lifecycle and engagement systems

I built production Snowflake and dbt models connecting attribution and Braze email, SMS, and push activity to downstream behavior and program milestones. During the migration, I found and corrected legacy logic, source, identity, and downstream measurement issues.

MDPP/D2C measurement architecture

For Lark’s Medicare DPP launch, I defined Marketing’s measurement path from paid source and campaign through enrollment, eligibility, account creation, activation, engagement, and billable milestones. I also built a downside, base, and upside planning model that shows CAC per account, cost per first billable event, likely funnel losses, and the assumptions that matter most.

Experiment design

For Samsung partnership measurement, I defined experiment-ready requirements that separate assignment from actual exposure and support ITT analysis, SRM checks, power planning, guardrails, and explicit decision rules.

I am also designing randomized geographic experiments for Google and Meta paid advertising. That work is in design and has not shipped.

Analyst development

I have coached other analysts at Lark alongside owning marketing analytics.