OPTIMIZE FOR CLICKS - BUHI SUPPLY CO. | STUKENT PROJECT

Project Goal:

This project was developed to enhance the optimization strategy used during the Buhi Supply Co. simulation by integrating Python-based scenario testing into the workflow. While Excel was initially used to analyze campaign performance and identify high-performing channels, this case study pushed the analysis further by leveraging Python’s data manipulation and visualization capabilities to test various budget allocation strategies, verify the accuracy of Excel’s recommendations, and uncover deeper trends.

The primary goal was to maximize total clicks within a fixed budget across multiple digital advertising channels, while building a scalable and automated framework for testing performance scenarios beyond static spreadsheets.

​

Solution:

• Data Setup & Exploration:

  • Loaded the “Optimize for Clicks” campaign dataset from the Stukent simulation, containing fields such as keyword, match type, bid, impressions, clicks, CTR, CPC, and quality score.

  • Cleaned and filtered the dataset using Python (Pandas) to remove incomplete or low-volume keywords, and focused the analysis on keywords with actionable volume and bid potential.

  • Explored correlations and trends across keyword attributes using Seaborn visualizations and z-score normalization to detect outliers in performance.

  • Segmented data by match type (exact, phrase, broad) and device type to understand differences in click behavior.

• Modeling Bidding Strategy:

  • Developed a bidding scenario model to test the impact of adjusted bids (or bid multipliers) across keyword match types and quality scores.

  • Used loop-based simulations in Python to iterate through various bid configurations (e.g., +10%, +20%, +30%) to estimate potential click gains and cost impact.

  • Compared performance across match types to evaluate which settings offered optimal clicks-per-dollar.

  • Conducted statistical testing to confirm non-linear trends (e.g., diminishing returns beyond a certain bid level).

• Keyword-level Optimization:

  • Identified low-performing keywords with low CTR, high CPC, or poor quality score and tested whether pausing, rebidding, or changing match type would yield better results.

  • Flagged exact match keywords with high search volume and moderate CTR as prime opportunities for higher bidding.

• Visualization & Insights:

  • Created plots showing relationships such as: bid vs click volume; match type vs CTR; cost per click vs quality score.

  • Built trade-off curves showing that while increasing bids boosted clicks, it also led to higher CPC and diminishing ROI, helping define optimal bid ceilings.

Results:

• Achieved 13% increase in total clicks compared to baseline (e.g. when using uniform bids) by optimizing keyword bids and match types.

• Lowered average cost per click (CPC) by 9% while maintaining or increasing click volume.

• Found that exact match keywords had 2.7x better CTR than broad match at similar bids.

• Identified a subset of 11 low-performing keywords that, when paused or re-bid, improved overall campaign ROI or reduced wasted spend.

• Verified the results by comparing Python-simulated optimizations with Excel’s Solver output, achieving consistent accuracy across tools.

​

Tools Used​:

• Stukent Simulation Platform: Campaign dataset (keywords, CTR, CPC, conversions)

• Excel

• Python (Google Colab)

• Pandas & Numpy

• Matplotlib & Seaborn

• Tableau