← Back to Portfolio
Project 03 — Digital Marketing Analytics

Maximise ROI: Analysing and Optimising Digital Marketing Campaigns

Multi-channel attribution · Markov chain modelling · A/B testing · Power BI

+27%
Projected ROAS improvement
+19%
Validated conversion uplift
140k
Sessions analysed
4
Channels remodelled
Python (pandas, NumPy)
Google Analytics
Power BI
Meta Ads Manager
Markov Attribution
A/B Testing
Statistical Significance
Project Narrative

Attribution, reallocation,
results.

Problem
Last-click attribution was systematically over-crediting paid search while starving profitable upper-funnel channels of budget. The business was optimising the wrong signal entirely.
Data
6 months of campaign data from Google Analytics, Meta Ads Manager, and an email platform. 140,000 sessions and 4,200 tracked conversions across four channels.
Solution
Data-driven multi-touch attribution model built in Python using Markov chain analysis. Each channel's true marginal contribution to conversion was isolated independently of the others.
Recommendation
Reallocate 22% of paid search spend to content marketing and email retargeting. Projected payback period of 6 weeks based on observed historical conversion rates per channel.
Impact
Simulated model projected a 27% improvement in blended ROAS. A controlled A/B test validated a 19% real-world conversion rate uplift, statistically significant at p < 0.05.
Method
Markov chain attribution, cohort comparison, two-tailed significance testing. Power BI executive dashboard built for senior stakeholder communication without technical jargon.
Data Visualisation

Attribution exposed.
Budget realigned.

Last-Click vs Markov Attribution
Conversion credit assigned per channel under each model
Budget Reallocation Recommendation
Spend distribution before and after optimisation
A/B Test Results: Conversion Rate Over Time
Control group (old allocation) vs Test group (reallocated budget) — 8-week observation window
ROAS by Channel (Before vs After)
Return on ad spend per euro invested, per channel
Conversion Funnel Drop-off
Session-to-conversion funnel analysis across 140k sessions
Tools & Methods
Python (pandas, NumPy) Google Analytics Meta Ads Manager Power BI Markov Chain Attribution A/B Testing Statistical Significance (p<0.05) Cohort Analysis ROAS Modelling

"The budget was optimised for the last click. The revenue was driven by what came before it."

← Back to Portfolio END OF PORTFOLIO