Data-driven product management is not just about tools or numbers — it’s about asking the right questions and turning data into decisions that move the needle.
Data science is no longer a niche skill reserved for specialized teams. It is now a fundamental part of your toolkit as a product manager. The actual job is to translate data into actionable decisions that create real value for users and the business.
Most product managers confuse data science with data reporting or dashboards. The trap is thinking that more data or prettier charts automatically mean better products. In practice, the skill lies in defining the right questions, designing experiments, and interpreting data with context.
This lesson draws on insights from Gaurav Bubna, a PM at Grab who leads AI and data-driven product initiatives. His experience at Grab, combined with Talvinder Singh’s coaching, illustrates how you can harness data science to sharpen your product intuition and execution.
Why data science matters for PMs
The tech industry in India and globally is undergoing a massive transformation powered by data. Whether you are a PM at a startup or a product leader in a large company, your ability to work effectively with data will determine your impact.
Talvinder explains: "If your company is not using data to build products, your competitors are. The companies that master data-driven product management pull ahead and redefine markets."
Data science helps you:
- Understand user behavior beyond gut feeling
- Validate hypotheses with experiments and metrics
- Prioritize product features based on evidence
- Predict outcomes and reduce risk
- Personalize user experiences at scale
Without these capabilities, you risk becoming a PM who manages opinions, not outcomes.
Core data science concepts for PMs
Gaurav emphasizes that PMs do not need to become data scientists but must grasp key concepts to collaborate effectively.
Define your objective clearly
Start by asking: What problem are you solving? What does success look like?
Talvinder: "Every data analysis starts with a question. You must know what you want to measure and why before looking at the data."
For example, if your goal is to increase user engagement, define how you will measure engagement. Is it daily active users, session length, or feature usage?
Understand your data and metrics
Data comes in many forms—event logs, user profiles, transaction records. You need to know:
- What data you have access to
- How reliable and clean the data is
- Which metrics truly reflect user value and business goals
Gaurav advises: "Focus on metrics that matter. Vanity metrics like total signups don’t tell you if users are actually benefiting."
Use hypothesis testing and experiments
To prove causality, you must conduct experiments such as A/B tests.
Talvinder references a classic case: "Obama’s 2008 campaign raised $60 million by running a simple experiment on their website. They tested different donation button texts and found the one that converted best."
This is the essence of data-driven product management—forming hypotheses, running tests, and iterating based on evidence.
Learn statistical concepts at a high level
You don’t have to master statistics but should understand concepts like:
- Null hypothesis
- P-values
- Confidence intervals
Talvinder: "Hypothesis testing formalizes what you already do instinctively: checking if a change made a real impact or if it was random chance."
Tools and technologies for PMs
The landscape of data tools is vast. Gaurav shares what he sees as essential for PMs in India:
| Tool Category | Examples | Use Case for PMs | Indian Context Notes |
|---|---|---|---|
| Data querying | MySQL, MongoDB | Extracting data for analysis | Relational and non-relational databases common in Indian startups |
| Data analysis | Excel, Google Sheets, R, Python | Basic to advanced data manipulation and modeling | Excel and Sheets are widely used; R and Python for deeper analysis |
| Data visualization | Tableau, Qliksense | Creating dashboards and reports | Tableau gaining traction in Indian enterprises |
| Analytics platforms | Mixpanel, Segment, Google Analytics | Tracking user events and funnels | Mixpanel popular with SaaS and consumer apps |
| Experimentation tools | Optimizely, Google Optimize | Running A/B tests and measuring impact | Limited adoption but growing awareness |
| AI/ML Platforms | Google Cloud AI, AWS SageMaker | Integrating AI insights and predictive models | Indian startups increasingly exploring these |
Talvinder cautions: "Knowing how to query databases is a fundamental skill for PMs. Without it, you rely on others and lose agility."
AI and machine learning in product management
Gaurav leads teams at Grab building AI-powered products using natural language processing and computer vision. His role involves translating AI capabilities into user value.
Talvinder highlights that AI is not magic; it must solve a specific user problem better than alternatives.
The PM’s role with AI
- Set acceptance criteria in user terms, not just model metrics
- Design feedback loops to improve models with user data
- Manage expectations about AI’s limitations and error rates
- Own the cost model for AI inference and scaling
For example, at Grab, AI helps personalize offers and detect fraud. But the PM ensures that AI outputs are timely, accurate enough, and integrated smoothly into user workflows.
Pitfalls to avoid
- Treating AI as a press release rather than a product feature
- Optimizing model accuracy without considering user experience
- Building proprietary models when API calls suffice
Talvinder: "Most Indian startups are better off using existing AI APIs than building custom models upfront. The moat is in data and workflow integration, not just the model."
Real-world example: Improving user onboarding
Imagine you are a PM at a consumer app struggling with new user drop-off.
You observe via Mixpanel that users take too long to complete onboarding steps.
Hypothesis: An intro video will educate users and reduce onboarding time.
You run an A/B test comparing users who see the video versus those who don’t.
Results show a statistically significant decrease in onboarding time and higher activation rates.
You iterate by refining the video content and measuring impact continuously.
This approach combines data science, experimentation, and product intuition.
Getting started with data science as a PM
Talvinder recommends a pragmatic path:
- Learn to query databases. Start with SQL basics to pull your own data.
- Master spreadsheet analysis. Google Sheets and Excel are your friends.
- Understand your product’s key metrics. Define what matters and why.
- Run simple experiments. Use tools like Google Optimize or Mixpanel.
- Collaborate closely with data scientists and engineers. Speak their language.
- Keep learning AI basics. Understand what models can and cannot do.
Supporting media: Video from the AMA session with Gaurav Bubna
Test yourself: Defining metrics for a new feature
You are the PM at a Series A fintech startup in Bangalore. Your team is launching a new loan eligibility calculator feature aimed at 500,000 users. The CEO wants you to report success after one month.
The call: Which metrics would you track to measure if the feature is successful? How would you design an experiment to validate its impact?
Your reasoning:
Where to go next
- Build your analytics foundation: Data-Driven Product Management
- Learn experimentation techniques: Running Effective A/B Tests
- Understand AI for PMs: AI Product Strategy
- Master SQL basics: SQL for Product Managers
- Explore product metrics: Metrics and KPIs
- See how Indian startups use data: Case Studies in Indian Product Management
PL alumni now work at Flipkart, Google, Razorpay, PhonePe, Swiggy, Amazon, Microsoft, and 30+ other companies.