BUHI Supply Co. — Click Optimization
Python EDA • Bid Strategy • Campaign Analytics
Click volume increase
13%
vs. inform bid baseline
CPC reduction
9%
While maintaining volume
Exact vs broad CTR
2.7x
At similar bid levels
Keyword pruned
11
Low-performing re-bid or paused
OVERVIEW
Extended the BUHI Supply Co. simulation by integrating Python-based scenario testing into the campaign optimization workflow. While Excel Solver was used to identify high-performing channels, this project pushed the analysis further — using Python to model bidding strategies, detect keyword-level inefficiencies, and verify Excel's recommendations with statistical testing and visualization.
The primary goal was to maximize total clicks within a fixed budget across multiple digital advertising channels, while building a scalable, automated framework for testing performance scenarios beyond static spreadsheets.
WHAT I BULT
Data setup & exploration
Loaded and cleaned the Stukent campaign dataset using Pandas. Explored correlations across keyword attributes with Seaborn, z-score normalization, and segmentation by match type and device.
Bidding scenario modeling
Built a loop-based simulation to test bid multipliers (+10%, +20%, +30%) across keyword match types and quality scores — estimating click gains and cost impact for each configuration.
Keyword-level optimization
Identified 11 low-performing keywords with poor CTR, high CPC, or low quality score. Tested whether pausing, rebidding, or changing match type would improve overall ROI.
Visualization & trade-off analysis
Built trade-off curves showing the relationship between bid increases, click volume, and CPC — defining optimal bid ceilings before diminishing returns set in.
OUTCOMES
-
Achieved 13% increase in total clicks vs. uniform bid baseline through keyword bid and match type optimization
-
Reduced average CPC by 9% while maintaining or increasing click volume
-
Exact match keywords delivered 2.7x better CTR than broad match at similar bid levels
-
Identified 11 low-performing keywords — pausing or rebidding improved campaign ROI and reduced wasted spend
-
Cross-validated Python simulation results against Excel Solver output — achieving consistent accuracy across both tools
KEY LEARNING
-
Bid adjustments and keyword match type have a meaningful, measurable impact on click efficiency — not just volume
-
High impressions with low CTR is a signal to prune, not boost — careful keyword pruning outperforms broad budget increases
-
Raising bids has a ceiling: click volume gains plateau while CPC keeps rising — trade-off curves make this ceiling visible
TOOLS USED
Full code and analysis on GitHub