A media performance intelligence case study exploring television advertising efficiency, attribution reliability, and network-level optimization through Python analytics and Tableau dashboards.
This project combines business intelligence, attribution analytics, KPI engineering, and executive-facing visualization to evaluate how television advertising performance varies across networks and over time. The analysis focuses on identifying which TV networks deliver scalable and cost-efficient customer acquisition while validating whether TV airings generate measurable short-term traffic impact.
Rather than relying solely on traditional survey attribution, the project integrates traffic-response analysis to evaluate whether TV exposure produces immediate website engagement beyond baseline traffic levels.
The final workflow combines:
- performance analytics
- attribution validation
- KPI engineering
- dashboard storytelling
- temporal trend analysis
- budget optimization strategy
- executive-facing business intelligence
Television advertising represents a major acquisition channel, but advertising efficiency varies substantially across networks.
Traditional attribution methods based on customer surveys often fail to fully capture TV’s short-term impact on online engagement. As a result, media buyers face two major challenges:
- determining which networks generate efficient customer acquisition
- validating whether TV airings actually drive measurable traffic lift
This project explores:
- Which networks deliver scalable and efficient performance?
- How does TV performance evolve over time?
- Do TV airings generate measurable short-term traffic impact?
- How can attribution confidence be improved?
- Where should advertising budget be reallocated?
The project followed a hybrid analytics and attribution workflow designed to combine operational KPI analysis with behavioral traffic validation.
Raw advertising & traffic data
→ Data cleaning & standardization
→ KPI engineering
→ Performance aggregation
→ Network-level analysis
→ Temporal trend analysis
→ Attribution traffic analysis
→ Dashboard development
→ Executive recommendations
The final analysis integrated:
- advertising spend data
- customer survey attribution
- TV airing schedules
- minute-level traffic behavior
- network-level performance metrics
Survey responses from customers who reported hearing about the product through television advertising.
Daily network-level spend and lift metrics used to evaluate overall TV performance efficiency.
Timestamp-level TV airing records used for attribution and traffic-response analysis.
Minute-level website traffic and baseline traffic estimates used to evaluate short-term post-airing traffic lift.
A major portion of the project focused on preparing fragmented advertising and attribution datasets into a unified analytical framework suitable for business intelligence reporting.
- Standardized inconsistent network naming
- Handled missing and “Other” network responses
- Removed incomplete or invalid observations
- Processed negative and zero-lift edge cases
- Unified timestamp and month formatting
- Prepared Tableau-ready aggregated datasets
- Created attribution-ready traffic alignment tables
The project engineered core advertising performance metrics including:
Spend / Impressions × 1000
Measures media delivery cost efficiency.
Spend / Lift
Measures the cost required to generate incremental traffic.
TV Attributable Purchases / Lift
Measures traffic conversion effectiveness.
Spend / TV Attributable Purchases
Measures customer acquisition efficiency.
The campaign demonstrated meaningful operating scale:
- Approximately $554K in total spend
- Approximately 8K attributable purchases
- Approximately $69 overall CPA
The analysis showed that TV advertising can drive measurable acquisition volume, but aggregated performance masks substantial network-level variation.
One of the strongest findings was the large performance disparity across networks.
Key findings included:
- Performance was concentrated within a small number of high-value networks
- Several high-spend networks delivered below-average efficiency
- Top-performing networks achieved acquisition costs less than half of underperforming networks
- Efficiency and scale did not consistently align
This suggested strong opportunities for strategic budget reallocation.
The analysis compared monthly performance patterns across the campaign period.
Key findings included:
- Spend increased from April to May
- Purchases increased alongside spend expansion
- CPA improved from April to May
- Conversion rates strengthened over time
- June data appeared operationally incomplete and unsuitable for optimization decisions
The results suggested that campaign efficiency improved as the campaign scaled.
A major component of the project focused on validating TV attribution using traffic-response analysis rather than relying exclusively on customer survey reporting.
TV airing timestamps were aligned with subsequent website traffic activity at the minute level.
The analysis estimated short-term incremental traffic lift immediately following TV airings in order to evaluate whether TV exposure produced measurable behavioral response.
Website traffic consistently increased immediately following TV airings.
Key observations included:
- Traffic spikes peaked within the first minute after airing
- Visitor counts exceeded baseline traffic levels
- TV exposure showed measurable short-term engagement effects
This provided strong directional evidence supporting TV attribution validity.
Traffic-response analysis also revealed substantial variation across networks.
Key findings included:
- A small subset of networks generated disproportionate traffic lift
- Traffic response consistency varied significantly within networks
- Some networks generated high spend with weak behavioral response
This reinforced the importance of network selection in overall media ROI optimization.
- Reallocate spend toward networks delivering both low CPA and strong traffic lift
- Reduce or pause investment in consistently inefficient networks
- Prioritize scalable high-efficiency channels
- Favor networks demonstrating stable performance across multiple airings
- Evaluate consistency rather than relying only on average performance
- Expand investment in high-performing network segments
- Combine survey attribution with traffic-response validation
- Exclude incomplete periods from optimization decisions
- Improve confidence through multiple attribution signals
Interactive Tableau dashboards were developed to support executive-facing business intelligence reporting.
- Overall campaign KPI tracking
- Monthly performance analysis
- Network-level efficiency comparison
- Attribution traffic visualization
- Scale vs efficiency analysis
- Temporal trend analysis
The dashboards were designed to support recurring campaign reporting and future optimization workflows.
- Python
- pandas
- NumPy
- matplotlib
- Tableau
- Interactive dashboards
- Executive KPI reporting
- Media attribution
- KPI engineering
- Traffic-response analysis
- Performance analytics
- Budget optimization
- Business intelligence
media-performance-intelligence/
│
├── analytics/ # Python analysis & KPI engineering
├── dashboards/ # Tableau packaged workbooks
├── visuals/ # Dashboard previews & analytical charts
├── reports/ # Executive reports & presentations
└── README.md
This repository is intentionally positioned as:
- a media attribution analytics case study
- a business intelligence & dashboarding project
- a TV advertising performance investigation
- a KPI engineering and optimization workflow
- an executive-facing analytics portfolio project
It is intentionally NOT positioned as:
- a generic Tableau assignment
- a dashboard-only visualization project
- a simple marketing analytics exercise
- a survey reporting workflow
- a classroom business intelligence project
- media-performance-intelligence
- attribution-analytics
- business-intelligence
- advertising-analytics
- dashboard-storytelling
- KPI-engineering
- budget-optimization
- TV-attribution
- performance-analytics
- executive-reporting
- network-efficiency-analysis
- traffic-response-analysis