A Practical Data Analysis Exercise for Product Managers
If you want to see data analysis in action, this post is for you
I am sure you have come across a ton of content that explains, at a high level, how to analyze data as a product manager and use these insights to make decisions. In this post, I wanted to take this a step further and do an actual, practical data analysis exercise. From this exercise, I will go through what questions I asked, what insights were pulled from the analysis, and finally identify possible initiatives based on those insights.
You can find this dataset on Kaggle. I’ve changed the dates to be more recent.
The goal of this post is to demystify data analysis for product managers by just doing it in public (warts and all). So let’s go.
Bike Rental Service Data Analysis Exercise
As the product manager for our bike rental service, I've plunged deep into our data from 2022 to 2023. This analysis has uncovered fascinating insights that will shape our strategy moving forward. I’ve posed five questions that I think can help us uncover useful insights.Â
1. How has the overall bike rental demand changed from 2022 to 2023?
The line plot comparing daily rentals from 2022 to 2023 tells a story of impressive growth. We see a significant increase in overall demand, with 2023 consistently outperforming 2022 across all seasons. This growth is particularly pronounced during the peak summer months, where we also see greater day-to-day variability. We can see this as well on the monthly chart.
Key insights:
- Our business experienced substantial growth from 2022 to 2023.
- Seasonal patterns remain consistent year-over-year.
- Increased variability in 2023 suggests we may be reaching capacity during peak times.
Potential initiatives:
To capitalize on this growth trend, we should consider expanding our bike fleet to accommodate the increasing demand, especially during peak seasons. Investing in infrastructure, such as docking stations and maintenance facilities, will be important to support this growth. We should also analyze the factors contributing to our growth and develop strategies to maintain this momentum. Implementing a yield management system could help us maximize revenue during high-demand periods while also managing capacity more effectively.
2. How does bike rental demand vary across different seasons and weather conditions?
Our heatmap reveals a compelling story about the interplay between seasons, weather, and bike rental demand. Summer emerges as our best season, with consistently high rental numbers across all weather conditions. The warm temperatures and longer daylight hours create the perfect environment for cycling, attracting both commuters and leisure riders.
Spring and fall show similar patterns, acting as transition seasons. During these months, clear weather becomes a crucial factor, significantly boosting rental numbers. This suggests that many of our users are fair-weather cyclists, ready to ride when conditions are favorable.
Winter tells a different story. The cold temperatures and shorter days lead to a noticeable dip in rentals, particularly during poor weather conditions. However, we see a small but dedicated group of riders braving the elements, especially on clear winter days.
Our bar chart visualizes this in another way and we can see again that the spring and summer seasons are our breadwinners.
Key insights:
- Seasonal variations significantly impact our business.
- Weather conditions play a crucial role in daily rental patterns.
- Summer is our peak season, while winter presents challenges.
Potential initiatives:
Based on these insights, we could implement dynamic pricing based on seasons and weather forecasts. This would allow us to maximize revenue during high-demand periods and incentivize ridership during slower times.
3. What are the peak hours for bike rentals, and how do they differ between weekdays and weekends?
Our heatmap and line plot of hourly rentals paints a vivid picture of urban life and leisure patterns. On weekdays, we see two distinct peaks that mirror the pulse of the city: a morning surge from 7-9 AM and an evening rush from 5-7 PM. These peaks align perfectly with typical commute times, suggesting that a significant portion of our weekday users are using our bikes as a means of getting to and from work.
The weekend pattern tells a different story. Instead of sharp peaks, we see a more distributed pattern of usage, with demand gradually rising from late morning and remaining high through the early evening. This suggests a shift from utilitarian commuting to leisure riding.
Key insights:
- Weekday peaks align with typical commute times, suggesting many users are commuters.
- Weekend usage indicates leisure riding throughout the day.
- Late-night demand is consistently low regardless of the day.
Potential initiatives:
To capitalize on these patterns, we could optimize bike availability and maintenance schedules around peak hours, ensuring maximum availability during high-demand periods. Introducing commuter-specific membership plans or bundles could cater to our weekday peak users. For weekends, we might create "day-trip" packages to encourage longer rentals and explore partnerships with local attractions.Â
4. How does the proportion of casual vs. registered users change throughout the year?
Our stacked bar chart reveals an intriguing dynamic between our casual and registered users. Throughout the year, registered users form the backbone of our business. However, the summer months bring a noticeable shift with a significant uptick in the proportion of casual users, likely due to increased tourism. We can see this summer uptick on the monthly chart as well.
The winter months see the lowest proportion of casual users, indicating that our winter riders are primarily committed, year-round cyclists. Encouragingly, we notice a slight increase in the overall proportion of registered users from 2022 to 2023.
Key insights:
- Our business relies heavily on registered users for consistent revenue.
- Seasonal tourism likely contributes to the increase in casual users during summer.
- We've been successful in converting some casual users to registered users over time.
Potential initiatives:
To leverage these insights, we could launch a targeted summer campaign to convert casual users to registered members, perhaps offering special conversion deals. Developing a loyalty program could help retain our valuable registered users and encourage year-round usage. Additionally, implementing a referral program for registered users could help bring in new customers and further grow our user base.
5. Is there a correlation between temperature and bike rental demand?
Our scatter plot and box plot reveal a predictable relationship between temperature and bike rental demand. We see a strong positive correlation up to about 70°F, after which demand plateaus decreases slightly at high temperatures (above 80°F). The highest median demand occurs in the 70-80°F range, suggesting this is the "sweet spot" for bike rentals.
Key insights:
- Temperature is a key factor in driving rental demand.
- Extreme heat may discourage some users from renting bikes.
- The "sweet spot" for bike rentals appears to be between 60-80°F.
Potential initiatives:
We could develop a predictive model for demand based on temperature forecasts, allowing us to optimize bike distribution. For very hot days, we might introduce heat-beating promotions, such as 2x loyalty points, to encourage ridership.
Takeaways
This deep dive into our data has revealed that our bike rental product and service is growing. We've uncovered the rhythms of our business – the daily pulse of commuters, the seasonal ebb and flow of casual users, and the overarching influence of weather and temperature. The potential initiatives discussed for each insight provide a potential roadmap for capitalizing on our strengths and addressing our challenges.
As we move forward, our focus should be on implementing these initiatives in a phased approach, carefully monitoring their impact, and continually refining our strategies. By leveraging these data-driven insights and remaining responsive to changing patterns, we can enhance user experience, drive growth, and solidify our position in the urban transportation landscape. Regular data analysis will continue to be our compass, guiding us toward new opportunities and helping us navigate the dynamic nature of our business.
I hope you found this exercise useful. This was meant to illustrate to you how the process of data analysis to potential initiative might go but it certainly isn’t comprehensive and the initiatives may very well just scratch the surface.
Great walkthrough. I think data analysis skills are underestimated for product managers. In fact, for some jobs I have been politely requested not to interfere with the job of the tech team. "They don't like people checking their data".