OPERATION ANALYTICS AND
INVESTIGATING METRIC SPIKE
BY ANANWITA SARKAR
Project Agenda
1. Project Description: The project's objective is to leverage operational analytics for end-to-end analysis of a
company's operations, identify areas for improvement, and provide actionable insights to cross-functional teams. As the
Data Analyst Lead at Microsoft, I will collaborate closely with departments like operations, support, and marketing to derive
valuable insights from their data. The primary focus is on optimizing workflows, enhancing automation, and predicting the
company's overall growth or decline. Additionally, the project will involve analyzing metric spikes in daily engagement and
sales, investigating their causes, and providing insights to address any dips in these metrics.
2. Approach: This project involves gathering data sets and tables from various departments and sources within the
company, ensuring data quality and normalization during the import into an SQL database. Through SQL queries and
aggregations, the collected data will be explored and analyzed to uncover patterns, trends, and anomalies. Key
performance metrics, such as daily engagement, sales, and customer satisfaction, will be identified in collaboration with
cross-functional teams and tracked using SQL queries. Metric spikes will be investigated using SQL subqueries, analyzing
historical data and collaborating with departments for contextual understanding. Insights and recommendations derived
from SQL analysis will drive process improvements, workflow optimization, and automation.
3. Tech-Stack Used: Used MySQL workbench 8.0 community server version 8.0.33 which is owned by oracle.
Task 1: Calculate the number of jobs reviewed per hour per day in November 2020
Output:
Insights: The number of jobs reviewed per hour per
day in November 2020 varies, with higher activity on
some days and lower activity on others.
Investigate if there were any technical issues or other
factors affecting user engagement.
Task 2: Calculate 7 day rolling average of throughput? For throughput, do you prefer
daily metric or 7-day rolling and why?
Output:
Insights: The 7-day rolling average of throughput provides
a smoothed view of the data, allowing you to observe trends
over time without being affected by daily fluctuations.
Continue using the 7-day rolling average for throughput
analysis, as it provides a more stable representation of
performance trends. This can help in identifying long-term
patterns and making more informed decisions.
Task 3: Calculate the percentage share of each language in the last 30 days?
Output:
Insights: The language distribution in the last 30 days is relatively
balanced, with Persian having the highest share.
Consider investing in language-specific content or features to enhance user
engagement in languages with lower shares.
Task 4: Display duplicate row count from the table?
Output:
Insights: There is one duplicate row in the data based on the
actor_id column.
Implement data validation mechanisms to prevent such duplicates
in the future.
Task 5: Calculate the weekly user engagement?
Output:
Insights: User engagement seems to have peaked around week
30 and has shown some fluctuations over the observed period.
Look for patterns related to content updates, marketing campaigns,
or any external events that might have influenced user behavior.
Use these insights to plan future engagement strategies.
Task 6: Calculate the user growth for product
Output:
Insights: User growth has generally been positive over time, with some fluctuations.
While user growth is positive, it's important to understand the factors driving this growth. Analyze the periods of higher growth and
see if they align with specific product updates, marketing efforts, or market trends. This can help in replicating successful strategies to
sustain and accelerate user growth.
Task 7: Calculate the weekly retention of users-sign up cohort?
Output:
Insights: Weekly user retention shows a gradual decline over time.
Focus on improving user retention strategies. Identify key touchpoints in
the user journey where users might be dropping off and work on
enhancing user experience, engagement, and value during those stages.
Task 8: Calculate the weekly engagement per device?
Output: *
Insights: Engagement varies across different devices and weeks. Some
devices consistently show higher engagement than others.
Consider optimizing the user experience for devices that show lower
engagement. Additionally, monitor device trends over time to adapt your
strategies and prioritize user engagement on devices with the highest potential.
* this is just a sample output of only 22 rows. There are 491 rows returned from the above
query which could not be accommodated in a single page.
Task 9: Calculate the email engagement metrics?
Output:
Insights: The email engagement metrics show an open rate of
approximately 33.58% and a click rate of about 14.79%.
Compare these metrics with industry benchmarks to determine how
your email engagement measures up. If your rates are below average,
consider refining your email content, subject lines, and targeting to
increase engagement. Regularly test and optimize your email campaigns
to improve these metrics.
Result:
Working on this project has allowed me to delve into the intricacies of operational analytics, gaining a
deeper understanding of its principles and methodologies. Through data integration, I have learned how to
effectively merge and normalize diverse datasets, ensuring accurate and reliable analysis. Employing SQL
analysis techniques has sharpened my ability to query, aggregate, and uncover insights from complex
datasets. Furthermore, collaborating closely with cross-functional teams has enhanced my communication
and teamwork skills, enabling me to effectively translate data findings into actionable recommendations for
process optimization, thus contributing to the overall success of the project.
THANK YOU
