Marketing Analytics - Muzz

A few days ago, I worked on a project for Muzz , an online halal dating company for Muslims. The objective was simple:

To find out what camaigns are working and where can we save money or invest to make the business more profitable and cater to user demands.

Two channels : Facebook and Tiktok Four Campaigns : Campaign A/ B /C /D (Anonymised)

Two countries : UK and France

End Goal : Make a pipeline where where we take raw datasets , process it and display on a pretty and functional dashboard for the marketing team.

First things first , to avoid you with the boring technical details , here's the end product, the dashboard is interactive , feel free to interact with it.

Now , we can proceed to the nitty gritty details:

Dataset

The project used 5 anonymised datasets:

1. facebook.csv – ad spend in GBP

2. tiktok.csv – ad spend in USD (converted to GBP)

3. fx_rates.csv – currency conversions for GBP/EUR

4. signups.csv – user registrations with campaign attribution

5. transactions.csv – customer payments in GBP/EUR

   
   

👉 View Dataset

Pipeline Steps

1. Ingestion: Imported all CSVs into BigQuery.

2. Cleaning & Transformation:

   • Removed duplicates and nulls

   • Converted currencies to GBP

   • Split campaign IDs and fixed missing FX weekends

   • Deduplicated sign-ups to keep only the first registration per account

3. Validation: dbt tests to ensure no nulls, duplicates, or invalid values slipped in.

4. Final Table: Built campaign_performance in BigQuery → the single source for all KPIs.
   

👉 View SQL/dbt Models (Empty Link for now , in the process of uploading for viewing purposes)

Dashboard

The campaign_performance table powers a Looker dashboard with:

• Funnel view: Spend → Sign-ups → Paying Customers

• Channel comparison: Facebook vs TikTok

• Country performance: UK vs France

• Campaign breakdown: A, B, C, D side-by-side
  

👉 View Dashboard 

Key Findings

• Funnel: Only ~14% of sign-ups became paying customers → major drop-off after sign-up.

• TikTok vs Facebook: TikTok outperformed, delivering cheaper customers and stronger returns.

• Campaigns:

  • Campaign D was the most efficient (every £1 spent returned £1.70).

  • Campaign A underperformed and lost money.

• Country: UK campaigns were more profitable; France lagged.

  
  

Recommendations

So with the data all transformed and insights popping out of the graphs , here are some recommendations I suggested :

• Increase spend on TikTok and Campaign D.

• Optimise or pause Campaign A.

• Reallocate part of the France budget to UK until performance improves.

• Add onboarding nudges (emails, offers, discounts) to reduce funnel drop-off.
  

Github Link