7.4 Agile Planning, Estimation, and Metrics
Key Takeaways
- User stories use the format "As a [role], I want [goal] so that [benefit]" and should satisfy the INVEST criteria
- Story points are RELATIVE measures of effort, complexity, and uncertainty — not hours — and cannot be compared across teams
- Velocity is the average story points completed per Sprint (average 3-5 Sprints) and forecasts how many Sprints remain — never used to compare teams
- Estimation techniques include Planning Poker (Fibonacci), T-shirt sizing, affinity estimation, and Wideband Delphi
- Burndown charts show remaining work, burnup charts show completed work plus a moving scope line, and a CFD reveals WIP and bottlenecks
How Agile Planning Differs
Agile planning is progressive (rolling-wave): instead of one detailed upfront schedule, teams plan at multiple horizons — product vision, release, and Sprint — and refine continuously as they learn. The Agile Planning Onion layers these from outermost (strategy/vision) inward to the Daily Scrum.
User Stories
A user story is a short, user-centered statement of need:
"As a [user role], I want [goal] so that [benefit]."
Examples:
- As a customer, I want to search products by category so that I can find items quickly.
- As an administrator, I want to export users to CSV so that I can report to management.
A story is the card that triggers a conversation and produces a confirmation (acceptance criteria) — Ron Jeffries' "3 C's."
INVEST criteria
| Letter | Criterion | Meaning |
|---|---|---|
| I | Independent | Deliverable on its own |
| N | Negotiable | Details discussed, not fixed |
| V | Valuable | Delivers user/business value |
| E | Estimable | Team can size it |
| S | Small | Fits in one iteration |
| T | Testable | Has clear acceptance criteria |
Backlog hierarchy
| Level | Size | Description |
|---|---|---|
| Theme | Largest | A grouping of related epics |
| Epic | Large | Too big for one Sprint; split into stories |
| User Story | Medium | Fits in one Sprint |
| Task | Smallest | A unit of work inside a story |
Estimation Techniques
Story points
Story points capture relative effort, complexity, and uncertainty — they are not hours. A 5-point story is harder than a 3-point story but not "five hours." Teams calibrate against a reference story and use a Fibonacci scale (1, 2, 3, 5, 8, 13, 21) or modified Fibonacci (…20, 40, 100). The gaps grow on purpose: large items carry too much uncertainty for fine precision. Points are team-specific — Team A's 8 means nothing to Team B.
Planning Poker
- Each estimator holds a Fibonacci card deck.
- The Product Owner reads a story; the team clarifies.
- Everyone reveals a card simultaneously (avoids anchoring).
- High and low voters explain; discuss.
- Re-vote until consensus.
Other techniques
| Technique | Best for |
|---|---|
| T-shirt sizing (XS-XL) | Fast, coarse sizing of big backlogs |
| Affinity estimation | Sorting many stories by relative size quickly |
| Dot voting | Quick prioritization |
| Wideband Delphi | Expert items needing discussion rounds |
Velocity
Velocity = average story points completed per Sprint, counting only items that meet the Definition of Done.
| Sprint | Completed |
|---|---|
| 1 | 20 |
| 2 | 24 |
| 3 | 18 |
| 4 | 22 |
| Average | 21 / Sprint |
If 126 points remain: 126 ÷ 21 = 6 Sprints to finish. Rules: average over 3–5 Sprints for stability; never compare velocity across teams (points differ) and never weaponize it to push teams faster — that corrupts the estimates.
Agile Charts (Information Radiators)
Burndown chart
- Y = work remaining, X = time. An ideal line trends to zero.
- Actual line above ideal = more work remaining than planned = behind.
- Actual below ideal = ahead. A flat actual line signals a blocked Sprint.
Burnup chart
- Y = work completed, X = time, plus a separate scope line.
- Its advantage over burndown: a rising scope line makes scope creep visible — a burndown can hide added scope as "no progress."
Cumulative Flow Diagram (CFD)
- Stacked bands show item counts in each stage over time.
- A widening band = a bottleneck (work entering faster than leaving).
- The horizontal width of a band approximates lead/cycle time; the vertical thickness approximates WIP.
Backlog Refinement and the Definition of Ready
Backlog refinement (grooming) is the ongoing add-detail/estimate/re-order activity. It is not a formal Scrum event and should consume no more than ~10% of Developer capacity. The optional Definition of Ready (DoR) is the team's checklist that a story is well-formed (meets INVEST, has acceptance criteria, is estimated, dependencies known) before it is pulled into Sprint Planning.
Relative vs. Absolute Estimation — Why Points Win
The exam wants you to understand why agile teams estimate relatively. Humans are poor at absolute estimates ("this will take 14.5 hours") but good at relative comparison ("this is about twice as hard as that one"). Story points exploit that strength, and because points absorb effort, complexity, and uncertainty in one number, a risky-but-small task can correctly out-point a large-but-routine one. A subtle trap: if a team converts points to hours every Sprint, they have abandoned the benefit and recreated time estimation.
Another trap is comparing two teams' velocities to judge productivity — because each team calibrates its own reference story, a velocity of 30 on one team and 20 on another tells you nothing about who delivers more value.
Reading the Charts Like an Examiner
Know the shape the chart should make, not just its axes. A healthy burndown trends steadily toward zero; a long flat plateau signals a blocking impediment, and a line that drops then rises means scope was added mid-Sprint. A burnup shows two lines converging — completed work climbing toward a scope line — and a stepped scope line is the unmistakable signature of scope creep that a burndown would hide. On a cumulative flow diagram, parallel bands of steady thickness mean smooth flow, while a band that fans out wider over time marks a stage where work arrives faster than it leaves — the bottleneck.
When a scenario describes a symptom ("progress stalled," "scope kept growing," "work piling up before testing"), map the symptom to the chart shape to pick the right diagnostic tool.
A team completed 22, 18, 26, and 22 story points over four Sprints. With 88 points remaining, how many more Sprints should they forecast?
Which statement about story points is TRUE?
In a Sprint burndown chart, the actual line is consistently ABOVE the ideal line. What does this indicate?
Why might a Product Owner prefer a burnup chart over a burndown chart?