1 Point (Very Small)

Example: text change, constant value update
Time: 1-2 hours
Complexity: very low
Risk: none

2 Points (Small)

Example: simple bug fix, add logging
Time: 2-4 hours
Complexity: low
Risk: low

3 Points (Medium)

Example: simple CRUD API endpoint
Time: 4-8 hours
Complexity: medium
Risk: low

5 Points (Medium-Large)

Example: complex form implementation, auth middleware
Time: 1-2 days
Complexity: medium
Risk: medium

8 Points (Large)

Example: new feature (frontend + backend)
Time: 2-3 days
Complexity: high
Risk: medium

13 Points (Very Large)

Example: payment system integration
Time: 1 week
Complexity: very high
Risk: high
Recommended: Split into smaller tasks

21+ Points (Epic)

Required: Must be split into smaller stories


### Step 2: Planning Poker

**Process**:
1. Product Owner explains the story
2. Team asks questions
3. Everyone picks a card (1, 2, 3, 5, 8, 13)
4. Reveal simultaneously
5. Explain highest/lowest scores
6. Re-vote
7. Reach consensus

**Example**:

Story: "Users can upload a profile photo"

Member A: 3 points (simple frontend) Member B: 5 points (image resizing needed) Member C: 8 points (S3 upload, security considerations)

Discussion:

Use an image processing library
S3 is already set up
File size validation needed

Re-vote → consensus on 5 points


### Step 3: T-Shirt Sizing (quick estimation)

```markdown
## T-Shirt sizes

- **XS**: 1-2 Story Points (within 1 hour)
- **S**: 2-3 Story Points (half day)
- **M**: 5 Story Points (1-2 days)
- **L**: 8 Story Points (1 week)
- **XL**: 13+ Story Points (needs splitting)

**When to use**:
- Initial backlog grooming
- Rough roadmap planning
- Quick prioritization

Step 4: Consider risk and uncertainty

Estimation adjustment:

interface TaskEstimate {
  baseEstimate: number;      // base estimate
  risk: 'low' | 'medium' | 'high';
  uncertainty: number;        // 0-1
  finalEstimate: number;      // adjusted estimate
}

function adjustEstimate(estimate: TaskEstimate): number {
  let buffer = 1.0;

  // risk buffer
  if (estimate.risk === 'medium') buffer *= 1.3;
  if (estimate.risk === 'high') buffer *= 1.5;

  // uncertainty buffer
  buffer *= (1 + estimate.uncertainty);

  return Math.ceil(estimate.baseEstimate * buffer);
}

// Example
const task = {
  baseEstimate: 5,
  risk: 'medium',
  uncertainty: 0.2  // 20% uncertainty
};

const final = adjustEstimate(task);  // 5 * 1.3 * 1.2 = 7.8 → 8 points

Output format

Estimation document template

## Task: [Task Name]

### Description
[work description]

### Acceptance Criteria
- [ ] Criterion 1
- [ ] Criterion 2
- [ ] Criterion 3

### Estimation
- **Story Points**: 5
- **T-Shirt Size**: M
- **Estimated Time**: 1-2 days

### Breakdown
- Frontend UI: 2 points
- API Endpoint: 2 points
- Testing: 1 point

### Risks
- Uncertain API response time (medium risk)
- External library dependency (low risk)

### Dependencies
- User authentication must be completed first

### Notes
- Need to discuss design with UX team

Constraints

Required rules (MUST)

Relative estimation: Relative complexity instead of absolute time
Team consensus: Agreement from the whole team, not individuals
Use historical data: Plan based on velocity

Prohibited (MUST NOT)

Pressuring individuals: Estimates are not promises
Overly granular estimation: Split anything 13+ points
Turning estimates into deadlines: estimate ≠ commitment

Best practices

Break Down: Split big work into smaller pieces
Reference Stories: Reference similar past work
Include buffer: Prepare for the unexpected

References

Metadata

Version

Current version: 1.0.0
Last updated: 2025-01-01
Compatible platforms: Claude, ChatGPT, Gemini

task-estimation

Task Estimation

When to use this skill

Instructions

Step 1: Story Points (relative estimation)