Marketing data environment, planning, and diagnosis

I worked on marketing planning, web performance, attribution, and the data environment behind executive reporting.

Role / scope
Senior Analyst, Marketing. August 2024 to November 2025
Timeframe
August 2024 to November 2025
Business problem
How should Q4 media be funded, what was driving a sharp decline in web traffic, and what data environment would make those answers trustworthy?
Key result
Recommended $6.9M in Q4 spend against an outside proposal of about $11M. Reduced manual reporting by at least 35%. Unified more than five marketing sources in Databricks.

Fragmented sources, delayed decisions

Marketing analytics was operating with fragmented data across five major siloed platforms. That blocked a trustworthy view of the marketing funnel and cross-channel ROI, and left more than 35% of analyst time on manual data wrangling instead of analysis.

Key initiatives such as the Leaderboard program and Summer Season Pass lacked a unified performance measurement framework. Leadership needed a Q4 media decision on a three-hour deadline, and web traffic was down about 42% year over year.

A Databricks-centered marketing data environment

I designed a Databricks-centered marketing-data environment integrating more than five sources, improved Adobe Analytics reporting, and established UTM and attribution governance.

Marketing data flow used for reporting and decision support

01

Sources

  • SFMC
  • Adobe Analytics
  • Google Ads / GA4
  • Meta Ads
  • Sprout Social

02

Databricks

  • ELT pipelines
  • Modeling and unification
  • Shared marketing metrics

03

Reporting and decisions

  • Executive reporting
  • Campaign ROI
  • Leaderboard and Season Pass measurement
  • Self-service in Power BI

Contribution modeling, root-cause analysis, and an MMM prototype

Under the Q4 deadline, I built a media-contribution model instead of accepting the external $11M proposal as the planning baseline. For the traffic decline, I traced a major driver to a 40% to 75% deterioration in CTR or message resonance, depending on the comparison, and presented the findings to VP leadership.

I also prototyped the company’s first internal Marketing Mix Model using Google Meridian, and selected and recommended a junior analyst candidate, then led his onboarding and analytical development.

Internal evidence under a hard clock

The Q4 decision could not wait for a fully mature cross-channel system. The contribution model had to be good enough to brief the CEO and CMO in hours, while the Databricks environment was being built to make later questions less dependent on manual wrangling.

Unifying sources and governing UTMs reduced reporting labor. The immediate executive question was still allocation. The internal model supported $6.9M. The outside proposal was about $11M.

Fund the internally evidenced plan, then keep measurement centralized

The $6.9M recommendation shaped CEO and CMO planning. The Databricks environment became the source of truth for marketing KPIs, executive reporting, and self-service analysis in Power BI.

Figure 1 · Q4 media recommendation

Internal contribution model versus outside proposal

Proposed Q4 media spend

Outside proposal

~$11M

Internal recommendation

$6.9M

Three-hour clockInternal modelCEO and CMO planning

$6.9M is the internally evidenced plan. About $11M is the outside proposal.

Less wrangling, a funded plan, and a usable funnel view

Manual reporting fell by at least 35%. More than five sources were unified, which enabled cross-channel ROI analysis and a consistent measurement frame for Leaderboard and Summer Season Pass.

The Q4 model recommended $6.9M against an outside proposal of about $11M. The web diagnosis identified weaker CTR and message resonance as an important driver of the 42% traffic decline.

All work