How Software Testing Works in Companies

QA in Practice · Inside Company Workflows

AK Ankit Kaushik · Lead Trainer, Data Engineering & AI | Updated 28 August 2026 | 18 min read Free · QA Guide

Company QA workflow — Stage 1: Pre-Dev Planning, Stage 2: Dev + Test, Stage 3: Release, then production + monitoring.

Quick summary — how software testing works in companies

Testing in companies is a system, not a step. This guide explains the real workflow — QA teams, test plans, release cycles, bug tracking, and CI/CD — from ticket to production.

In this guide you will learn:

Stage 1: Pre-Dev Planning — requirements and test plans.
Stage 2: Dev + Test — manual + automation running together.
Stage 3: Release — sign-off, regression, and CI/CD.
How QA teams are structured — roles and ownership.
Metrics companies track — coverage, bugs, escaped bugs.

SECTION 01Stage 1: Pre-Dev Planning

Testing in a company doesn't start with code — it starts with planning. Here's the breakdown:

Step	Focus	QA Actions	Deliverable
Step 1	Read Requirements	Review user stories, acceptance criteria, and edge cases	Questions + clarifications
Step 2	Test Plan Creation	Define scope, test types, tools, and schedule	Test plan document
Step 3	Test Case Design	Write test cases, including edge cases	Test case repository
Step 4	Environment Setup	Prepare test data, staging, and test accounts	Ready test environment

Pre-Dev Planning — How It Works:

1. QA joins the requirements meeting.
   - Reads the user story and acceptance criteria.
   - Asks: "How can this break?"

2. QA writes a test plan.
   - Scope (in-scope, out-of-scope)
   - Test types (functional, regression, security)
   - Tools (pytest, Playwright, Postman)
   - Timeline + resources

3. QA writes test cases.
   - One test case per scenario.
   - Includes edge cases and negative paths.

4. QA sets up environments.
   - Staging environment ready.
   - Test data prepared.
   - Test accounts created.

Why this matters:
- QA finds ambiguity before code exists.
- Saves 10x rework later.
- "Shifting left" is a real thing.

Test Plan Template (one page):

Feature: ____________
Owner: ____________
Release: ____________

Scope:
- In-scope: ______
- Out-of-scope: ______

Test Types:
□ Functional
□ Regression
□ Security
□ Accessibility
□ Performance

Environments:
- Staging URL: ______
- Test data location: ______

Test Cases:
- #1: ______
- #2: ______
- ...

Risks:
- ______
- ______

Sign-off Criteria:
- All critical tests passing
- Zero critical bugs
- Regression suite passing

QA Lead signature: ____________
Date: ____________

stage1-planning.md

Key insight: The most expensive bugs are the ones baked into requirements. QA's job in Stage 1 is to find them before code exists.

SECTION 02Stage 2: Dev + Test

This is where most QA work happens — testing while development is ongoing. Here's the breakdown:

Step	Focus	QA Actions	Deliverable
Step 1	Feature Testing	Execute test cases as features are built	Bug reports + test results
Step 2	Automation	Write/update automated tests for stable features	Automated test suite
Step 3	API Testing	Test endpoints directly with Postman or requests	API test results
Step 4	Bug Reporting	Log bugs with steps, severity, and evidence	Jira tickets

Dev + Test — How It Works:

1. Feature is ready in staging.
   - QA is notified via Slack/Jira.
   - QA runs test cases.

2. Manual testing first (usually).
   - Happy path, edge cases, error states.
   - QA logs bugs found.

3. Automation covers the stable parts.
   - After a feature is stable, QA writes
     automated tests.
   - Automation runs on every PR.

4. API testing in parallel.
   - Test endpoints with Postman.
   - Verify auth, status codes, error handling.

5. Bug reporting.
   - QA logs bugs with severity + steps.
   - Developer fixes.
   - QA retests and closes.

Tools used:
- Jira, Linear, GitHub Issues (tracking)
- Playwright, Selenium, Cypress (UI)
- Postman, pytest, REST Assured (API)
- Slack, Teams (communication)

Bug Report Template:

Title: [Component] Short description
Example: [Login] Password reset fails for +91 numbers

Environment:
- URL: staging.example.com
- OS/Browser: Windows 11 / Chrome 120
- Build: v2.4.1-rc3

Severity: High
Priority: P1

Steps to reproduce:
1. Open login page
2. Click "Forgot password"
3. Enter +919876543210
4. Click Submit

Expected:
Reset email should arrive within 2 minutes.

Actual:
No email received after 30 minutes.
Server log shows: "SMS provider error 503"

Evidence:
- Screenshot of error screen
- Log snippet attached
- Video of reproduction

Assigned to: @dev-arjun
Labels: bug, authentication, P1

stage2-dev-test.md

Key insight: The best bug reports save developers hours — title, steps, expected, actual, and evidence. A vague "login broken" ticket wastes everyone's time.

SECTION 03Stage 3: Release & Sign-off

Release is where QA gives the final green light. Here's the breakdown:

Step	Focus	QA Actions	Deliverable
Step 1	Regression Testing	Run the full regression suite	Regression report
Step 2	Smoke Testing	Quick sanity check of core flows	Smoke pass/fail
Step 3	Sign-off	QA Lead approves release	Sign-off document
Step 4	Post-Release Monitoring	Watch logs, errors, user reports	Incident triage

Release & Sign-off — How It Works:

1. Release candidate is built.
   - Usually from main branch.
   - Deployed to a pre-prod environment.

2. Regression suite runs.
   - Full test suite on the release candidate.
   - If any critical test fails, release is blocked.

3. Smoke testing on production.
   - After deployment, quick sanity checks.
   - Catch post-deploy issues in minutes.

4. QA Lead signs off.
   - Checklist signed: regression, smoke, critical bugs.
   - Go/no-go decision is documented.

5. Post-release monitoring.
   - Watch error rates, latency, user complaints.
   - Any spike → rollback plan.

Roles:
- Dev: builds the release.
- QA Lead: signs off.
- Ops/DevOps: deploys and monitors.

CI/CD Pipeline Example:

On every pull request:
- Lint
- Unit tests (fast, ~2 minutes)
- Integration tests (~5 minutes)

On merge to main:
- Full test suite (unit + integration + API)
- Build artifact
- Deploy to staging

On release tag:
- Full regression suite
- Deploy to pre-prod
- QA sign-off gate
- Deploy to production

Automated gates that block merges:
□ Any failing test
□ Coverage drop below threshold
□ Security scan failure

Manual gates:
□ QA Lead sign-off for major releases
□ Product owner approval for pricing changes

stage3-release.md

Key insight: Release is a shared responsibility — QA doesn't own production quality alone. Developers, QA, and Ops all have gates and responsibilities.

SECTION 04QA team structures

Here are the common ways QA teams are structured in companies:

QA Team Models:

1. Centralised QA team
   - QA reports to a QA manager.
   - Testers work across multiple product teams.
   - Pros: consistency, shared tools.
   - Cons: can feel disconnected from product.

2. Embedded QA (most common in 2026)
   - One QA per product team (or two).
   - Sits with dev + product.
   - Pros: deep context, faster decisions.
   - Cons: less cross-team consistency.

3. Hybrid
   - Embedded QA + central QA leadership.
   - Central team sets standards and tools.
   - Embedded team owns product testing.
   - Common in mid-size to large companies.

4. No dedicated QA (rare)
   - Developers own testing.
   - Only works with strong testing culture.
   - Common in very early startups.

Roles & Ownership:

QA Engineer (manual):
- Test plans, test cases, execution
- Bug reporting
- Some API testing
- Communicates with dev + product

QA Automation Engineer:
- Writes automated tests
- Maintains test suites
- Owns CI integration
- Debugs flaky tests

SDET (Software Development Engineer in Test):
- Writes production-quality test code
- Builds test frameworks
- Often works inside dev team
- Reviews code and design

QA Lead / QA Manager:
- Owns QA strategy + roadmap
- Reviews sign-offs
- Coordinates with leadership
- Handles QA hiring and growth

Product Owner (interacts with QA):
- Defines acceptance criteria
- Reviews QA findings
- Owns release decisions

Skills by Role:

QA Engineer (manual):
- Test design
- Bug reporting
- Basic SQL
- API testing basics
- Domain knowledge

QA Automation Engineer:
- Programming (Python/JS/Java)
- Test frameworks (pytest, Playwright)
- CI/CD
- Git
- Debugging

SDET:
- All automation skills +
- System design
- Framework architecture
- Code review
- Performance testing basics

QA Lead:
- All above +
- Strategy + planning
- Communication
- Mentoring
- Metrics and reporting

qa-team-structures.md

SECTION 05Metrics companies track

Here are the QA metrics companies actually track:

Test Coverage — % of code covered by automated tests. Aim for 70-80% on critical paths, not 100% everywhere.
Escaped Bugs — bugs that reached production. Lower is better. This is the most honest QA metric.
Defect Density — bugs per feature or per 1000 lines of code. Tracks product quality over time.
Cycle Time — how long a bug takes from report to fix. Fast cycle times keep QA efficient.
Test Pass Rate — % of tests passing in CI. Volatile pass rate = flaky tests = wasted time.

Pro tip: Vanity metrics like "bugs found" reward the wrong behaviour. Real QA metrics measure whether bugs reach users — escaped bugs, MTTR (mean time to repair), and user-reported issues.

SECTION 06Interview Q&A — QA in companies

Q1When does QA actually start in a company's release cycle?

In mature companies, QA starts at requirements review — before any code is written. This is called "shift-left" testing. Testers who only test after code is deployed are behind modern practice.

Q2Do QA teams work in sprints?

Yes — most companies use Agile and QA works inside the sprint alongside developers. Testing happens continuously within the sprint, not as a separate "testing phase" at the end.

Q3Who signs off before a release goes to production?

Usually the QA Lead signs off on quality, but shared ownership is common. Developers confirm changes, product owners approve the scope, and QA confirms all tests pass. It's a collective gate, not a single signature.

Q4How do companies measure QA success?

Not by counting bugs found — that rewards late-stage testing. Companies measure escaped bugs (bugs that reached production), MTTR, and user-reported issues. The best QA teams reduce incidents over time.

Q5Is there still manual testing in companies?

Yes — but less than before. Manual testing is still important for exploratory, usability, and edge cases. But most repetitive testing is automated, and pure-manual QA roles are shrinking.

SECTION 07Test yourself — QA in companies quiz

Five questions. No sign-up.

0 / 5

Pick an answer to see why it is right or wrong.

SECTION 08Frequently asked questions

When does QA start in the release cycle?

Ideally at requirements review — before code is written. Modern companies practice "shift-left" testing, where QA reads requirements and writes test cases before any feature is built.

Do companies use manual or automated testing?

Both. Manual testing is used for exploratory, usability, and edge cases. Automation covers repetitive cases (regression, API, smoke tests). Most companies aim for 70-80% automation on stable paths.

What tools do QA teams use?

Jira or Linear for tracking, pytest/Playwright/Selenium for automation, Postman for API testing, GitHub Actions or GitLab CI for pipelines. Source code lives in Git, and CI runs tests on every PR.

How is QA structured in big vs. small companies?

Small companies often have 1-2 QA engineers embedded in a product team. Large companies usually have a hybrid model — embedded QA + a central QA leadership team that owns standards, tooling, and strategy.

Is QA respected in modern companies?

Yes — but the role has changed. Pure manual QA is often undervalued. Automation engineers and SDETs are highly respected because they combine coding and testing. Modern companies treat quality as a shared responsibility.

What I would do in the first 30 daysLayoff · 14 min Why sending 500 resumes won't workJob Search · 14 min Rebuild your resume for the AI job marketResume · 15 min Which skills to learn before applyingSkills · 14 min 90-day career recovery roadmapRecovery · 16 min How AI can help you prepare for interviewsAI · 11 min

Classroom & online · Noida

Learn QA in the way companies actually hire for it

Our Automation Testing Course is designed to help you build real skills — pytest, Playwright, API testing, and CI/CD pipelines used by real QA teams.

₹15,500 · full programme ~~₹24,000~~

Explore the course Download the syllabus

Hands-on automation projects
Real CI/CD pipelines
QA team workflow training
Weekday & weekend batches

How Software Testing Works in Companies