Product analyst

What does a product analyst do?

A product analyst uses data on how people use a product to help a team decide what to build, fix or drop.

The product can be an app, a website or an internal tool. You measure funnels, retention and the results of experiments.

In Singapore, product analysts work in banks, fintechs, e-commerce firms, software companies, telcos and government agencies. You usually sit in a product, growth or data team. Some report to a head of product, others to a head of data.

You work daily with product managers, designers and engineers. You also work with marketers, data engineers and, where personal data is involved, compliance or data protection staff.

What the work involves

• Write SQL queries to pull usage data and check that the numbers are reliable.
• Build and read funnels, retention curves and cohort reports in a tool such as Amplitude or Mixpanel.
• Plan A/B tests with product managers, then analyse the results and recommend whether to ship.
• Update dashboards in Looker Studio or a similar tool for weekly product reviews.
• Present findings to the product team and suggest what to try next.

Pay in Singapore

Back to top

Skills a product analyst needs

Back to top

Product analyst job description

We are hiring a product analyst to join our product team in Singapore.

You will study how customers use our app and website, run experiments and turn the findings into clear recommendations. You will work closely with product managers, designers and engineers. You should be comfortable with SQL, statistics and explaining results to people who do not work with data.

Responsibilities

• Define the metrics that show whether a product or feature is working.
• Check that tracking events are set up correctly and report gaps to engineers.
• Query and clean data from databases and product analytics tools.
• Analyse funnels, retention and user segments to find where customers drop off.
• Design A/B tests with product managers, including sample size, run time and success measures.
• Analyse experiment results and recommend whether to ship, change or stop a feature.
• Build and maintain dashboards for product teams and senior managers.
• Present findings and next steps in simple language to non-technical colleagues.
• Follow company rules on personal data when you collect and share analysis.

Requirements

• A degree in statistics, mathematics, economics, computing, business analytics or a related field is common. Employers also consider strong portfolios and work experience.
• Usually 1 to 3 years in analytics, product or growth work for a mid-level role. Junior roles may accept fresh graduates with projects.
• Confident SQL, including joins, window functions and cohort queries.
• Working knowledge of statistics, including hypothesis testing and confidence intervals.
• Experience with a product analytics tool such as Amplitude or Mixpanel, a dashboard tool such as Looker Studio, and basic Python or R.

Back to top

Product analyst career path

Most product analysts start in a junior analyst role and move up through senior and lead roles.

The most senior role, Head of Product Analytics, sets analytics strategy for the organisation.

1. Junior product analyst

   Entry

   You pull data with SQL, update dashboards and support experiments under guidance from a senior colleague.

2. Product analyst

   Mid

   You own the metrics for a product area, run experiments yourself and present results to product managers.

3. Senior product analyst

   Senior

   You lead analysis on large product decisions, set experiment standards and coach junior analysts.

4. Product analytics manager

   Lead

   You manage a team of analysts, agree priorities with product leaders and set how insights are reported.

5. Head of Product Analytics

   Head · SkillsFuture title

   You set the data strategy for product, build the business case for improvements and guide research and testing across the organisation.

Steps marked "SkillsFuture title" follow the career pathway in the SkillsFuture Skills Framework. The others are titles commonly used in Singapore.

Related jobs and their pay

Back to top

Product analyst interview questions

We wrote these questions from the skills this job needs. They are not an official question bank.

1. Write a SQL query to find 7-day retention by signup week from a users table and an events table.

   What a good answer covers

   • Defines the cohort by signup week and counts its size.
   • Joins events to users and keeps events 1 to 7 days after signup.
   • Counts distinct users instead of events.
   • Divides returning users by cohort size.
   • Notes that dates should use Singapore time (UTC+8) if the business reports in local time.

2. How would you design an A/B test for a new checkout button on a Singapore e-commerce app?

   What a good answer covers

   • States a clear hypothesis and one primary metric, such as checkout completion rate.
   • Adds guardrail metrics such as refund rate or page load time.
   • Picks the randomisation unit, usually the user rather than the session.
   • Estimates sample size from baseline rate and the smallest change worth detecting.
   • Runs the test over full weeks to cover weekday and weekend behaviour.

3. A product manager asks what a p-value of 0.04 means. How do you explain it?

   What a good answer covers

   • Says it is the chance of seeing a result at least this extreme if there were no real effect.
   • Makes clear it is a different thing from the chance the feature works.
   • Mentions effect size and confidence intervals as well as significance.
   • Warns about checking results many times and stopping early.
   • Uses simple words and an example.

4. Your experiment shows 52% of users in one group and 48% in the other, but you planned a 50/50 split. What do you do?

   What a good answer covers

   • Recognises a possible sample ratio mismatch.
   • Runs a statistical check, such as a chi-square test, on the split.
   • Looks for causes such as bugs in assignment, bot traffic or logging gaps.
   • Holds back the result until the cause is found.
   • Tells engineers and the product manager what was found.

5. A funnel in Amplitude shows a sharp drop at step three after a release. How do you check whether it is real?

   What a good answer covers

   • Checks whether the tracking event changed or broke in the release.
   • Compares the drop across platforms, app versions and user segments.
   • Cross-checks with a SQL query on raw data or backend records.
   • Looks at related metrics, such as completed orders.
   • Reports confidence level and next steps to engineers.

6. When would you use Python instead of SQL for an analysis? Give an example.

   What a good answer covers

   • Uses SQL for joins, filtering and aggregation close to the data.
   • Uses Python for statistical tests, modelling and repeated or complex steps.
   • Gives an example, such as bootstrapped confidence intervals for an experiment.
   • Mentions pandas or a similar library for cleaning and reshaping data.
   • Keeps code readable and reusable so others can run it.

7. What would you put on a weekly product dashboard in Looker Studio for a mobile banking app?

   What a good answer covers

   • Picks a small set of metrics tied to product goals, such as active users, activation and retention.
   • Shows trends over time with comparisons to the previous period.
   • Adds filters for platform and user segment.
   • Defines each metric clearly on the page.
   • Checks the data source refreshes on schedule and flags when it fails.

8. How would you define success metrics for a food delivery app feature that recommends restaurants?

   What a good answer covers

   • Starts from the goal of the feature instead of the available data.
   • Chooses a main metric, such as orders from recommended restaurants.
   • Adds guardrails, such as order cancellations and delivery time.
   • Separates short-term and long-term effects.
   • Explains how the metric will be tracked and tested.

9. How would you use ChatGPT in your analysis work without putting company data at risk?

   What a good answer covers

   • Uses only tools approved by the employer.
   • Keeps personal data and confidential figures out of prompts.
   • Uses it to draft or explain SQL and code, then tests the result against known numbers.
   • Checks any statistical claims it makes before relying on them.
   • Records how the output was checked.

1. A test shows a higher conversion rate but lower seven-day retention. What do you recommend?

   What a good answer covers

   • Checks both results for statistical reliability.
   • Looks at which user segments drive each change.
   • Considers long-term value as well as conversion.
   • Suggests a follow-up test or a partial rollout.
   • Sets out the trade-off clearly so the product manager can decide.

2. A product manager wants to ship a feature after two days of a test that looks positive. How do you respond?

   What a good answer covers

   • Explains the risk of stopping early and false positives.
   • Points to the planned sample size and run time.
   • Shows what the current confidence interval looks like.
   • Offers options, such as a staged rollout with monitoring.
   • Stays respectful and focuses on the decision, not on winning the point.

3. Weekly active users fall 10% on a Monday. How do you find the cause?

   What a good answer covers

   • Checks for tracking or pipeline problems first.
   • Splits the drop by platform, region, app version and acquisition channel.
   • Checks for outages, releases, campaigns or public holidays.
   • Compares with the same week in earlier periods.
   • Shares a short update early and a full explanation once confirmed.

1. Tell us about a time your analysis changed a product decision.

   What a good answer covers

   • Describes the decision and the original plan.
   • Explains what data was used and how it was checked.
   • Shows how the findings were presented to stakeholders.
   • States the outcome and what was measured afterwards.
   • Is honest about what the analysis could not show.

2. Describe a time you had to explain a difficult finding to someone who disagreed with it.

   What a good answer covers

   • Listens to the objection and checks it against the data.
   • Uses simple charts and everyday language.
   • Admits uncertainty where it exists.
   • Finds common ground, such as a follow-up test.
   • Keeps the working relationship intact.

1. Your team wants to track detailed user behaviour in an app used in Singapore. What do you check under the Personal Data Protection Act (PDPA)?

   What a good answer covers

   • Identifies which events and fields count as personal data.
   • Checks that users have given consent and been told the purpose.
   • Collects only what the analysis needs.
   • Uses pseudonymised IDs and limits who can see raw data.
   • Asks the company data protection officer when unsure.

Common questions about a product analyst job

Sources