A case study analyzing smart device usage data to generate marketing recommendations for Bellabeat, a wellness technology company for women.
Analyze non-Bellabeat smart device usage data to identify consumer behavior trends, then translate those trends into actionable marketing recommendations for the Bellabeat app.
Guiding questions:
- What are the trends in smart device usage?
- How could these trends apply to Bellabeat customers?
- How could these trends help influence Bellabeat marketing strategy?
R · RStudio · tidyverse · ggplot2 · lubridate · janitor · R Markdown
FitBit Fitness Tracker Data by Mobius (CC0 Public Domain) — 33 users, 31 days, April–May 2016.
| # | Finding |
|---|---|
| 1 | 51% of users are Sedentary or Lightly Active — average user logs only 7,638 daily steps |
| 2 | Activity peaks at 6 PM (599 avg steps), with a secondary peak at 12–2 PM |
| 3 | Sunday is the laziest day — 15% fewer steps than Saturday |
| 4 | Steps and calories show a moderate-to-strong correlation (r = 0.59) |
| 5 | Users spend 16.5 sedentary hours per day on average |
| 6 | Users spend 39 minutes nightly in bed but not asleep |
| 7 | 88% of users tracked their device for 21+ days — high commitment |
Target the realistic majority — busy users building daily wellness habits — rather than the fitness elite. Promote intensity-based metrics alongside step counts.
Schedule push notifications at 11 AM and 4 PM — right before users' natural activity peaks. Build a dedicated Sunday Reset campaign to address the weekly low.
88% of users tracked consistently — churn isn't the problem. Reinvest budget into onboarding conversion, hourly stand reminders, mindfulness wind-down campaigns, and premium membership upsell.
See /visualizations for all seven charts.
bellabeat-case-study/
├── data/
│ ├── raw/ # Original CSVs from Kaggle
│ └── clean/ # Processed datasets (7 files)
├── notebooks/ # R Markdown analysis files
│ ├── 01_process.Rmd # Data cleaning & transformation
│ ├── 02_analyze.Rmd # Statistical analysis & findings
│ └── 03_share.Rmd # Visualizations
├── visualizations/ # 7 exported charts (PNG, 300 DPI)
├── docs/ # Phase-by-phase documentation
│ ├── 01_ask.md # Business task & stakeholders
│ ├── 02_prepare.md # Data source & ROCCC assessment
│ ├── 03_process.md # Cleaning summary
│ └── 06_act.md # Recommendations
└── README.md # This file
Following the Google Data Analytics six-phase framework:
| Phase | Documentation |
|---|---|
| Ask | docs/01_ask.md — Business task & stakeholders |
| Prepare | docs/02_prepare.md — Data source & credibility |
| Process | docs/03_process.md — Cleaning & transformation |
| Analyze | notebooks/02_analyze.Rmd — Statistical findings |
| Share | notebooks/03_share.Rmd — Visualizations |
| Act | docs/06_act.md — Recommendations |
This analysis is directional, not definitive. The dataset has known limitations:
- Small sample — only 33 users
- Outdated — data collected in 2016
- No demographic information — notably no gender data, despite Bellabeat targeting women specifically
- Self-selected participants — MTurk users may not represent the general population
- Sleep gap — only 24 of 33 users logged sleep data
These limitations are addressed in detail in the Prepare documentation.
To reproduce this analysis:
- Clone this repository
- Download the dataset from
Kaggle
and place the CSV files in
data/raw/ - Open
bellabeat-case-study.Rprojin RStudio - Install required packages:
tidyverse,lubridate,janitor,skimr,scales - Knit the R Markdown files in order:
01_process.Rmd→02_analyze.Rmd→03_share.Rmd
Ahmad Daniel Aspiring Data Analyst · Google Data Analytics Certificate
📧 danieljasme5@gmail.com 💼 LinkedIn 🌐 Portfolio
Completed as the capstone project for the Google Data Analytics Professional Certificate on Coursera.