Skip to content
Mykhailo Polishchuk
← Back to projects
Marketing AnalyticsAttributionDashboarding

Marketing Mix Lab for Student Startup

Created a channel attribution and budget scenario model to improve CAC efficiency across paid social, email, and referral.

Published Nov 2025

Marketing mix dashboard with channel contribution bars

Overview (Problem + Context)

  • The student startup was increasing spend, but leadership lacked clarity on marginal channel performance.
  • Weekly reporting emphasized vanity metrics and did not support reallocation decisions.
  • The team needed a transparent framework to compare channel efficiency under multiple budget scenarios.

My Role + Responsibilities

  • Owned KPI definition and taxonomy for acquisition, activation, and cost performance.
  • Implemented consistent tracking conventions across paid and owned channels.
  • Facilitated review sessions to translate dashboard outputs into action.

Approach

  • Unified campaign and conversion data into a shared reporting model.
  • Built scenario assumptions for spend elasticity, conversion lag, and channel saturation.
  • Evaluated budget shifts against CAC and payback constraints.
  • Defined an operating cadence for weekly experimentation decisions.

Implementation Details (Stack + Architecture Choices)

  • Used SQL transformations to normalize source data from analytics and ad platforms.
  • Prototyped scenario engine in Python and exposed controls in Streamlit.
  • Delivered executive-facing views in Tableau for recurring growth meetings.
  • Structured dashboard sections by decision type: scale, hold, or reduce.

Results (Metrics + Outcomes)

  • Achieved a 24% reduction in blended CAC over a six-week period.
  • Reallocated budget toward high-intent channels with lower payback period.
  • Improved confidence in planning by replacing ad-hoc spreadsheets with standardized reporting.

What I’d Do Next (Reflection)

  • Add MMM-style baseline decomposition to separate seasonality from campaign impact.
  • Integrate retention and LTV estimates to support full-funnel budgeting.
  • Automate anomaly alerts to detect underperforming campaigns earlier.

Links

  • [GitHub repository](https://github.com/your-github/marketing-mix-lab)
  • [Dashboard walkthrough](https://example.com/marketing-mix-dashboard)
  • [Experiment report](https://example.com/marketing-mix-report)