Companies / Google

Google practice pack

Product company. Coding-first testing: clean code, edge cases, and testing at very large scale.

These are our own practice questions for the kind of QA/SDET role this company hires for. They are not the company's real interview questions. Confirm the current process on the company's official careers page.

Skills this role is usually tested on

  • Coding
  • Test design from first principles
  • Large-scale and flaky-test thinking
  • Debugging
  • Clear communication

Business areas used for scenarios

  • search
  • cloud
  • advertising
  • mobile platforms

6 questions

Answer out loud first, then open the guide. Most people read the guide too early.

  1. 1

    How would you test a search box with autocomplete?

    mediumTest design

    A strong answer covers

    Input classes (empty, long, unicode, emoji), latency budget, ordering and ranking stability, caching, privacy of suggestions, accessibility, behaviour under slow networks, and how you would measure quality, not just correctness.

  2. 2

    Write a function that merges overlapping intervals and list the tests you would add.

    mediumCoding

    A strong answer covers

    Sort by start, sweep and merge. Tests: empty, single, touching intervals, nested, unsorted, duplicates, negative values, very large input.

  3. 3

    A test fails about 1 in 200 runs across thousands of tests. How do you find the cause at scale?

    hardFlakiness

    A strong answer covers

    Collect per-test pass history, correlate with machine, time, order and shared resources, rerun in isolation many times, quarantine with an owner and deadline, and fix the common causes in shared helpers.

  4. 4

    Users report a feature is slow, but your tests show it is fast. What next?

    mediumDebugging

    A strong answer covers

    Your environment differs: data volume, region, device, network. Reproduce with realistic data, use field metrics and traces, and compare percentiles, not the mean.

  5. 5

    Design a test strategy for a feature that changes how results are ranked.

    hardDesign

    A strong answer covers

    Offline evaluation sets, regression on known queries, A/B experiment with guardrail metrics, human review samples, rollback plan, and monitoring for slow quality drift.

  6. 6

    You disagree with a developer about whether an issue blocks launch. How do you decide?

    easyCommunication

    A strong answer covers

    Frame it by user impact and risk, bring evidence, propose options (fix, mitigate, launch with a flag), and agree who makes the call and when.

Now say it to an interviewer

Reading a guide is not the same as answering under a clock. Run a mock interview and get scored feedback.

Start a mock interview Build a plan from a job description