Bellabeat Case Study

Understanding Smart Device Usage to Inform Bellabeat's Marketing Strategy

Project Overview

This case study was completed as my capstone project for the Google Data Analytics Professional Certificate and was my first end-to-end data analytics case study.

Working from the perspective of a junior data analyst for Bellabeat, a women-focused health and wellness technology company, I analyzed publicly available fitness and wellness data to identify patterns in smart-device use and translate those findings into practical marketing and engagement recommendations.

The business task was to answer three questions:

• What patterns appear in smart-device and wellness data?

• How might these patterns apply to Bellabeat's customers and products?

• How could Bellabeat use these insights to guide its marketing and user-engagement strategy?

Data and Analytical Approach

The primary analysis used publicly available Fitbit Fitness Tracker Data from Kaggle/Mobius. After combining the March–April and April–May 2016 activity files, the analyzed dataset contained 35 unique user IDs. Activity, sleep, and weight-tracking data were cleaned in Google Sheets and analyzed in BigQuery using SQL.

Because the Fitbit data did not contain demographic information, I also used the Sleep Health and Lifestyle dataset, a supplementary Kaggle dataset containing 374 records with variables such as gender, sleep duration, sleep quality, physical activity, stress, occupation, and BMI.

The supplementary dataset is synthetic/compiled rather than directly collected, so results from it are treated as exploratory illustrations rather than evidence about Bellabeat's actual customers.

Key Findings

The analysis identified several useful patterns.

• Weekly behavior followed a recognizable rhythm. Sunday had the lowest average step count at approximately 6,607 steps while also having the highest average sleep duration at approximately 7.55 hours. Saturday had the highest average step count at approximately 7,752 steps, while Thursday had the lowest average sleep duration at approximately 6.69 hours.

• Sedentary time had a stronger relationship with sleep duration than daily step count. Among records where activity and sleep data could be matched, sedentary minutes had a moderate negative correlation with sleep duration (r = -0.53), compared with a much weaker relationship between steps and sleep (r = -0.19).

• Stress and sleep showed the strongest relationship in the supplementary dataset. Stress had a very strong negative correlation with sleep quality (r = -0.90) and sleep duration (r = -0.81), while its relationship with daily steps was weak (r = 0.19). Because this finding comes from a synthetic supplementary dataset, it is best viewed as a hypothesis to validate using real Bellabeat customer data.

• Descriptive gender differences appeared in sleep and stress rather than activity. Average physical activity levels were nearly identical between women and men in the supplementary dataset, while women reported higher average sleep quality and lower average stress. Again, these results are exploratory because the underlying dataset is synthetic.

• Use of secondary tracking features was inconsistent. Only 24 of the 35 Fitbit activity users appeared in the available sleep dataset, and only 13 logged weight data. Users who did log weight also averaged more daily steps and less sedentary time, suggesting that broader engagement with tracking features may be associated with greater overall platform engagement.

Recommendations

Based on these findings, I developed three marketing and product-engagement recommendations for Bellabeat:

1. Emphasize sleep, stress management, and broader wellness rather than competing primarily on step counting.

2. Use day-specific notifications and content that respond to recurring weekly behavior patterns.

3. Improve onboarding and personalized engagement so users are introduced to multiple Bellabeat features rather than using only basic activity tracking.

These recommendations are intended as starting points for testing rather than definitive prescriptions. The most important next step would be validating them with current, first-party Bellabeat customer data.

Project Outcome

This project strengthened my skills in Google Sheets, BigQuery SQL, data cleaning, correlation analysis, data visualization, and business communication. More importantly, it introduced me to the importance of distinguishing an interesting analytical relationship from a conclusion that the available data cannot fully support.

The final presentation is embedded below. Additional pages document the project's detailed findings, recommendations, data-cleaning process, and SQL analysis.

Based on these findings, I developed three marketing recommendations for Bellabeat which consisted of emphasizing marketing around stress and sleep management rather than activity tracking alone, developing day-specific notifications and engagement strategies, and incorporating onboarding that encourages multi-feature use. 

This project strengthened my skills in SQL, spreadsheet analysis, data visualization, and translating analytical findings into actionable recommendations.

Please see the final presentation inserted below. For more details related to Key Findings and Recommendations or Documentation and Processes, please follow the links here or navigate to the dropdown menu in the top right hand corner under Bellabeat Case Study.