SQL + Tableau Ad Campaign Analysis: From Dirty Data to Insightful Dashboards #

1. Why I Built This #

I wanted to build something real-world and business-relevant, something that went beyond technical skill and showed I could tell a story with data. Marketing spend is a perfect topic for that.

Companies often dump money into ads without really knowing which campaigns are working. So I created a full workflow that starts with messy marketing data and ends with a clean, actionable dashboard that shows which campaigns actually delivered results.

2. How I Built It #

I started with a raw CSV of ad campaign data. To make the project more realistic (and more impressive), I simulated common data issues like:

• Typos and inconsistent text
• Null values
• Duplicates
• Outliers

The Stack: #

• SQL for cleaning, exploration, and building insights
• Python for previewing the data
• Tableau for creating an interactive story-driven dashboard

I broke the SQL work into modular scripts:

• inject_dirty_data.sql
• cleaning_steps.sql
• exploration_steps.sql
• analysis_queries.sql

Then I brought the clean data into Tableau and built a full KPI section, bar charts, scatter plots, and boxplots that told the story of which campaigns performed — and which didn’t.

3. Trade-Offs and Decisions #

Product per Dollar vs Z-Score Efficiency #

Originally I tried using a simple Products Sold per $1 Spent metric. It was interpretable, sure but it didn’t tell the whole story.

That’s why I created a custom Efficiency Index using z-scores, which centered performance around the average and let me easily see who over or underperformed.

What Counts as “Efficient”? #

I categorized campaigns as “Efficient” or “Inefficient” based on whether they scored above or below average. It made for cleaner visuals and simpler insights without sacrificing nuance.

4. What I Took Away from This #

This project helped me level up in a few major ways:

• SQL cleaning: I now feel confident dealing with messy data in SQL, even creating my own “dirty” dataset to prove I can handle it.
• Data storytelling: Tableau made it easy to explore patterns, but I learned how to turn those into insights especially when comparing spend tier performance.
• Realism matters: Injecting dirty data made the project feel authentic, not canned. It also let me show I can spot and fix issues that would break dashboards in the real world.

5. The Dashboard and the Data #

If you’re curious to explore it all visually:

• Check out the Tableau dashboard
• See the full project on GitHub
• Read the structured project summary

6. Let’s Connect #

If you’ve got thoughts, want to collaborate, or just like clean data and good dashboards, feel free to connect:

LinkedIn