Direct Mail Attribution Remapping & Channel Optimization

Project Overview
Upwork's direct mail program had been over-attributing performance due to flawed matching logic in its address-based conversion tracking. I led a forensic audit and redesign of this system, remapping the attribution logic to increase accuracy, transparency, and strategic accountability.
The result: a major budget shift away from inflated channels toward higher-performing and more measurable media investments.
Business Challenge
- Conversion attribution relied on loose 10-character regex logic and a 30-day lookback
- Caused false-positive matches and overreported results (734 vs. actual ~70)
- Created confusion around DM channel effectiveness
- Stakeholders lacked confidence in media performance data
Solution
🔍 Attribution Logic Redesign
- Replaced partial match rules with precise full-address regex logic
- Used deterministic match rules validated against source-of-truth systems
- Resolved address conflicts (e.g., Suite 302 vs. Suite 480) with clean attribution precision
📊 Historical Attribution Reassessment
- Backtested logic against campaigns from 2023–2024
- Discovered true conversions were just 82 (2023) and 67 (2024)
- Original reports had claimed 734 conversions—an 88% inflation
💡 Strategic Reallocation
- Presented findings to CMO and senior stakeholders
- Helped shift budget from direct mail into:
- Podcast advertising
- Billboard placements (Palo Alto)
- Paid media expansion
Conclusion
This engagement not only corrected a critical attribution flaw but also enabled Upwork to confidently reallocate $3M in budget toward high-impact channels. The case highlights the importance of precision in marketing analytics and how rigorous data governance supports smarter, leaner media decisions.
Related Project

Frequently Asked Questions


Fresh Takes From My Real-time Industry Experience
Our services help you create digital products Stay ahead of the curve with our latest insights, tips, and industry trends.
Let’s Build your
Growth Engine Together

