Scrum is built on empirical process control — three pillars: Transparency, Inspection, and Adaptation. It organises work into fixed-length iterations called Sprints (max. 4 weeks), each delivering a potentially shippable increment.
- Develops and communicates the product goal
- Owns and prioritises the Product Backlog
- Determines release dates and content
- Accepts or rejects completed work
- Ensures the Scrum process is followed
- Removes impediments for the team
- Facilitates Scrum events (time-boxing)
- Keeps Scrum artifacts up to date
- Responsible for Sprint Backlog items
- Ensures quality — upholds Definition of Done
- Keeps the Sprint Goal in focus
- Self-organising, cross-functional
- Usually max. 10 members
- No sub-teams — clear roles only
- Self-organising — decides how to do the work
- Cross-functional — all skills needed are present
- PO → maximises product value
- SM → maximises team effectiveness
- Developers → deliver the Increment
- All share responsibility for Scrum process
- Single — one backlog per product
- Transparent — visible to all
- Current — continuously refined
- Ordered — PO orders by value
- Contains the Product Goal and all Product Backlog Items
- Contains the Sprint Goal
- Selected items from the Product Backlog
- Execution plan for Developers
- Large items broken into smaller tasks
- Visualised on a task board: open / in progress / done
- One or more increments per Sprint
- Must meet Definition of Done
- Usable regardless of whether the PO releases it
- Each increment builds on all prior ones
A shared understanding of what "complete" means for any Increment. Creates transparency and ensures consistent quality across the team.
- If an item doesn't meet DoD → back to Product Backlog
- Cannot be presented at Sprint Review without meeting DoD
- Developers are required to conform to it
- If org-wide standard exists → all teams must follow it as minimum
- If no standard → Scrum Team creates their own DoD
- Multiple teams on same product → must share one DoD
Backlog items are structured from broad themes down to concrete, testable stories.
- Requirements are fixed during a Sprint
- No breaks between Sprints
- Max. 4 weeks; shorter is often better
- Contains all 4 recurring meetings
- Key tool for discipline and focus
- Scrum Master is responsible for enforcement
- Limits scale with Sprint duration (shown below)
- Prevents scope creep within the event
| Event | Participants | Time-box (1-month Sprint) | Purpose |
|---|---|---|---|
| Sprint | Entire Scrum Team | ≤ 4 weeks | Container for all other events; produces an Increment |
| Sprint Planning | Dev + PO + SM | max. 8 h | Select Product Backlog Items → Sprint Backlog; set Sprint Goal; plan execution |
| Daily Scrum | Dev + SM | max. 15 min | Sync on progress since last Daily, plan until next, surface impediments |
| Sprint Review | Scrum Team + stakeholders | max. 4 h | Demo Increment to PO & client; discuss and adapt Product Backlog |
| Sprint Retrospective | Team + SM | max. 3 h | Inspect team process; identify improvements; commit to changes next Sprint |
Each story must specify a distinct target group ("who"), a concrete functionality ("what"), and the value delivered ("why"). Examples:
- Makes the Definition of Done concrete per story
- Enables objective testing of completion
- Agreed between PO and Developers during refinement
- Drives the Sprint Review demo
- Testable — yes or no, not "mostly"
- Specific — concrete numbers or behaviours
- Agreed — not imposed unilaterally
- Minimal — only what's needed, not a full spec
- Compare backlog items relative to each other
- Not an absolute unit (not hours or days)
- Velocity = story points completed per Sprint
- Enables burn-down charts to visualise progress
- Teams calibrate around a shared reference story
- Velocity: stabilises after a few Sprints — use for forecasting
- Burn-down chart: remaining story points over time
- Ideal line vs. actual — shows if team is on track
- Velocity is team-specific — don't compare across teams
- Improves predictability over time
A consensus-based estimation technique. Each team member independently estimates, then all reveal simultaneously — divergence triggers discussion.
- Team selects a medium-sized reference story first
- PO reads out the story to be estimated
- Each member picks a card face-down
- All reveal simultaneously
- Extreme values discuss their reasoning
- New round until consensus
- Prevents anchoring bias from early reveals
- Everyone must commit to an independent estimate
- Forces structured discussion on outliers
- The process surfaces hidden complexity
Card values (modified Fibonacci — gaps reflect estimation uncertainty at scale):
Artifacts: Product Backlog, Sprint Backlog, Increment.
Events: Sprint, Sprint Planning, Daily Scrum, Sprint Review, Sprint Retrospective.
As [who] I want [what] so that [why].INVEST: Independent, Negotiable, Valuable, Estimatable, Small, Testable. These criteria ensure stories are well-formed, deliverable and verifiable.
Simultaneous reveal prevents anchoring bias — if one person reveals first, others tend to adjust toward that value rather than giving an independent estimate.
Feature — concrete functionality within that theme.
User Story — specific value for a target user group, written in the "As / I want / so that" format.
Task — small implementation steps broken out of a User Story, tracked in the Sprint Backlog.
Sprint Planning: max. 8 hours.
Daily Scrum: max. 15 minutes.
Sprint Review: max. 4 hours.
Sprint Retrospective: max. 3 hours.
For shorter Sprints, all time-boxes scale proportionally. The Scrum Master is responsible for enforcing them.