Context

We are a newly employed data analyst team in the Customer Experience (CX) at Vanguard, the US-based investment management company. We've been thrown straight into the deep end with our first task. Before our arrival, the team launched an exciting digital experiment, and now, they're eagerly waiting to uncover the results and need our help.

The Digital Challenge

The digital world is evolving, and so are Vanguard’s clients. Vanguard believed that a more intuitive and modern User Interface (UI), coupled with timely in-context prompts (cues, messages, hints, or instructions provided to users directly within the context of their current task or action), could make the online process smoother for clients. The critical question was: Would these changes encourage more clients to complete the process

• 🛠 GitHub Repository: vanguard-ab-test
• 📋 Trello Board: Project Tasks & Timeline
• Presentation: https://docs.google.com/presentation/d/1mef54hLp55CzS6i1yTwkKkXWRaOcK_bGKkalwjcDdv4/edit?slide=id.p1#slide=id.p1

Good UI increases conversion rates by 200%

Good UX can push it to 400%

Reason:

Vanguard is a global asset manager with demanding international customers that need 24/7 easy access to our platform. New digital customer expectation > New UI Method? A/B Test: Control: Old UI Test: New UI

Hypothesis: The new UI design will have a significantly better:

Completion Rate (>5%) Time Spend Error rates Engagement

3 Datasets : Client Profiles, Digital Footprints, Experiment Roster.

3/15/2017 to 6/20/2017

Type of information : Demographics, Digitals Behaviours, Account Activity.

|Completion Rate | Rejected| |Error Rate | Approved| |Duration for each steps | Approved| |Engagement | Rejected|

vanguard-ab-test/
├── data/                        # Raw and cleaned CSV files
├── notebooks/                   # Python notebooks with analysis
├── slides/                      # Presentation
├── README.md                    # Introduction of the purpose of this project
├── .gitignore                   # 
├── requirements.txt             # List of libraries and versions used 
└── requirements-dev.txt         # List of libraries and versions used, without Jupyter

• Python (Pandas, Seaborn, Matplotlib)
• Jupyter Notebooks
• Git & GitHub
• Trello (for task planning)
• Visual Studio Code IDE & plugins
• Tableau
• Canva

* Adam Askari, Guilherme Haas, Julie Tring, Levin Schily *