Key takeaway
- How testing works in Scrum: events, who tests, the four agile testing quadrants, a Definition of Done checklist, challenges and the role of AI agents.
What is Scrum testing methodology?
Scrum testing is not a separate method. It is the practice of building quality checks into every Sprint so that each Increment is verified before it is called done. The 2020 Scrum Guide defines Scrum as a framework and does not use the words "tester" or "testing" at all. It makes the Developers accountable for creating a usable Increment each Sprint, and it asks the team to meet a shared Definition of Done. Testing in Scrum is how a team earns that definition.
For a CTO or head of engineering, the practical question is who tests what, and when. This guide answers it with the Scrum events, the four agile testing quadrants, a Definition of Done you can audit, and the limits of the approach. For the wider discipline, see shift-left testing and continuous testing.
How does testing fit into the Scrum events?
Scrum has five events: the Sprint, Sprint Planning, the Daily Scrum, the Sprint Review and the Sprint Retrospective. A Sprint is a fixed-length event of one month or less (2020 Scrum Guide). Testing is not a phase at the end of it. It runs through every event.
| Scrum event | Testing contribution | Decision it supports |
|---|---|---|
| Sprint Planning | Clarify acceptance criteria, size test effort, agree what "done" needs for each item | What can the team verify within this Sprint? |
| Daily Scrum (15 minutes, for the Developers) | Surface blocked tests, failing builds and unverified work | Does the Sprint Goal still look achievable? |
| Sprint work | Write and run tests alongside the code; fix defects in the same Sprint | Is the Increment meeting the Definition of Done? |
| Sprint Review | Demonstrate the verified Increment and collect stakeholder feedback | Does it solve the right problem? |
| Sprint Retrospective | Review escaped defects, flaky tests and automation gaps | Which testing practice do we change next Sprint? |
The Scrum Guide gives Sprint Planning a maximum of eight hours for a one-month Sprint, and the Daily Scrum is 15 minutes. Both are timeboxes, so test planning has to be lightweight and continuous rather than a separate document-heavy activity.
Who is responsible for testing in a Scrum team?
Scrum has three accountabilities: Developers, the Product Owner and the Scrum Master. There is no separate tester accountability in the Scrum Guide. Developers are "the people in the Scrum Team that are committed to creating any aspect of a usable Increment each Sprint", and that includes verification. In practice many teams still have people who specialise in testing, but they work as part of the Developers, not as a downstream gate.
| Accountability | Role in quality | Watch out for |
|---|---|---|
| Developers (including testing specialists) | Build, test and verify the Increment; keep the tests current | Treating testers as a separate hand-off stage |
| Product Owner | Orders the Product Backlog and clarifies what acceptable behaviour means | Vague stories that cannot be tested |
| Scrum Master | Helps the team improve its practices and remove impediments | Reporting test counts instead of removing blockers |
How do the agile testing quadrants map to Scrum?
The agile testing quadrants, described by Lisa Crispin and Janet Gregory in their work on agile testing, sort tests along two axes: technology-facing or business-facing, and whether they guide development or critique the product. The numbering is arbitrary. According to the authors, it does not imply an order, and teams use techniques from all four quadrants concurrently in small increments.
| Quadrant | Focus | Typical tests | Who usually leads |
|---|---|---|---|
| Q1 | Technology-facing, guides development | Unit and component tests, connectivity checks | Developers |
| Q2 | Business-facing, guides development | Story tests written first, prototypes, simulations, UX checks | Product Owner with Developers |
| Q3 | Business-facing, critiques the product | Exploratory, usability and user acceptance testing, business-level workflows | Testing specialists, users and stakeholders |
| Q4 | Technology-facing, critiques the product | Performance, load and security testing, other quality attributes | Specialists with platform and security teams |
Source: Janet Gregory, Chapter 8 on the agile testing quadrants (read 13 November 2026). Many teams automate Q1 and much of Q2, and keep Q3 largely human. For the related testing types, see system testing and user acceptance testing.
What should a Scrum Definition of Done include for testing?
The Definition of Done is the commitment attached to the Increment. The Scrum Guide calls it a formal description of the state of the Increment when it meets the quality measures required for the product. A testing-aware version is specific enough to check. A starting point to adapt:
- Code reviewed and unit tests passing in the build pipeline.
- Acceptance criteria for the story demonstrated and agreed with the Product Owner.
- Regression suite green; no known defects left open without a decision.
- Non-functional checks agreed for the story (performance, security) completed or explicitly deferred with a reason.
- Test evidence recorded where audit or regulation requires it.
If a team cannot meet its Definition of Done, the Scrum Guide says the item returns to the Product Backlog rather than being released. That rule is what keeps the testing commitment honest.
What does an agile tester need to do differently?
Compared with a waterfall test phase, the work changes in four ways.
| Area | Traditional test phase | Scrum team |
|---|---|---|
| Timing | After development is complete | Throughout the Sprint, from refinement onward |
| Automation | Optional or late | Needed from the start so regression does not grow each Sprint |
| Collaboration | Hand-offs and defect tickets | Shared work with Developers and the Product Owner |
| Feedback | At release | At each Sprint Review |
Automation is the main skill shift. Unit and behaviour-driven frameworks such as JUnit and Cucumber let a team re-run regression every Sprint without growing the test effort. Start with unit testing and extend to API testing.
What are the common challenges and when does Scrum testing not fit?
| Challenge | Why it happens | Response |
|---|---|---|
| Short Sprints leave little time to test | Testing is left to the last days | Slice stories smaller; test continuously; automate regression |
| Changing scope mid-Sprint | Stakeholder requests arrive late | Route changes through the Product Owner and the Product Backlog |
| Testing seen as a separate team | Old structures carried into Scrum | Make verification a Developer-level commitment |
| Hard-to-test non-functional needs | Q4 work needs environments and data | Plan the environment early; run Q4 checks in the pipeline |
| Heavily regulated release evidence | Audit needs documented proof | Record evidence as a by-product of the pipeline |
Scrum is a weaker fit for work that cannot be split into small verifiable increments, such as a single hardware certification event or a fixed-scope migration with one cutover. In those cases a hybrid plan, with Scrum for the build and a staged test and release gate, can be more honest than forcing every check into a Sprint.
How are AI agents changing testing in Scrum teams?
AI agents can now draft test cases from user stories, propose regression selections and help triage failures. They change who does the repetitive work, not who is accountable. Under the Scrum Guide the Developers remain accountable for the Increment, so agent output should be reviewed, versioned and covered by the same Definition of Done. Governed context, meaning the policies, acceptance criteria and decision history an agent should use, matters more than the agent itself. Read agentic software testing for controls, and agentic workflows for patterns.
For teams building this capability, ElixirClaw is an agentic OS platform to build, orchestrate and govern AI agents and automation workflows while keeping humans in control. AgentQA is described on its site as automated quality assurance for AI, and ElixirData Context OS keeps governed context, versioned policy and decision traces in one decision layer above AI agents.
How should an enterprise adopt Scrum testing?
- Write a Definition of Done that includes testing and make it visible.
- Automate Q1 first, then the stable parts of Q2.
- Add exploratory and acceptance testing (Q3) to each Sprint Review cycle.
- Move Q4 checks into the delivery pipeline.
- Use the Retrospective to track escaped defects and flaky tests, then adjust.
- Introduce AI agents for drafting and triage only under human review.
For the continuous-delivery side, see continuous testing in DevOps.
To map one testing decision your agents should make, and the context and policy it depends on, book a Context Workshop. See what happens in the workshop.
Frequently asked questions
Is Scrum testing a separate methodology?
No. Scrum is a framework, and the 2020 Scrum Guide does not mention testers or testing. Testing is part of how the Developers verify each Increment and meet the Definition of Done.
Does Scrum have a tester role?
No. Scrum has three accountabilities: Developers, Product Owner and Scrum Master. Testing specialists work as part of the Developers.
What are the agile testing quadrants?
A taxonomy by Crispin and Gregory that sorts tests as technology-facing or business-facing, and as guiding development or critiquing the product. The numbering implies no order.
How long is a Scrum Sprint, and what does that mean for testing?
A Sprint is one month or less according to the 2020 Scrum Guide. Testing and automation must therefore run inside the Sprint, not after it.
What if an item fails the Definition of Done?
It is not released or presented as done. The Scrum Guide says it returns to the Product Backlog for future consideration.
Can AI agents replace testers in Scrum?
They can draft tests and help triage failures, but the Developers stay accountable for the Increment, so agent output needs human review.
Explore the XenonStack platform family: ElixirData · ElixirClaw · ElixirHub · NexaStack · MetaSecure · Neural AI · AgentGRC · AgentSRE · AgentAnalyst · AgentInstruct · AgentRAI · AgentScrum · AgentSearch · AgentForce · AI Aviator · SAIF Aviator · CohortIQ · XenonStack AI · XenonStack UAE