Three-Point Estimation Guide

Overview

Three-point estimation uses optimistic, pessimistic, and most likely estimates to calculate expected effort and assess uncertainty.

Basic Formulas

PERT Formula (Program Evaluation and Review Technique) 

Expected Estimate = (O + 4M + P) / 6

Where:

  • O = Optimistic estimate (best case)
  • M = Most likely estimate (realistic case)
  • P = Pessimistic estimate (worst case)

Triangular Distribution 

Expected Estimate = (O + M + P) / 3

Standard Deviation 

Standard Deviation = (P - O) / 6

Confidence Intervals

  • 68% confidence: Expected ± 1 Standard Deviation
  • 95% confidence: Expected ± 2 Standard Deviations
  • 99.7% confidence: Expected ± 3 Standard Deviations

Step-by-Step Process

Step 1: Define Scenarios

| Scenario | Probability | Conditions | |———-|————-|————| | Optimistic | ~10% | Everything goes perfectly | | Most Likely | ~60% | Normal development conditions | | Pessimistic | ~10% | Major problems occur |

Step 2: Gather Three Estimates

For each work package or feature:

  1. Optimistic (O): Minimal time if everything goes perfectly
  2. Most Likely (M): Realistic estimate under normal conditions
  3. Pessimistic (P): Maximum time if significant problems occur

Step 3: Calculate Expected Values

Apply PERT or triangular distribution formula

Step 4: Calculate Uncertainty

Determine standard deviation and confidence intervals

Practical Examples

Example 1: User Authentication Feature

  • Optimistic: 3 days (perfect conditions)
  • Most Likely: 5 days (normal development)
  • Pessimistic: 10 days (major security issues)

PERT Calculation:

Expected = (3 + 4×5 + 10) / 6 = 33/6 = 5.5 days
Standard Deviation = (10 - 3) / 6 = 1.17 days

Confidence Intervals:

  • 68% confidence: 5.5 ± 1.17 = 4.33 to 6.67 days
  • 95% confidence: 5.5 ± 2.34 = 3.16 to 7.84 days

Example 2: Database Integration

  • Optimistic: 8 hours
  • Most Likely: 16 hours
  • Pessimistic: 40 hours

PERT Calculation:

Expected = (8 + 4×16 + 40) / 6 = 112/6 = 18.67 hours
Standard Deviation = (40 - 8) / 6 = 5.33 hours

Guidelines for Estimating

Optimistic Estimate (O)

  • Assume perfect conditions
  • No interruptions or delays
  • All dependencies available
  • Team fully focused
  • No rework required

Most Likely Estimate (M)

  • Normal working conditions
  • Typical interruptions
  • Some minor issues to resolve
  • Average team productivity
  • Minimal rework

Pessimistic Estimate (P)

  • Significant problems occur
  • Key team members unavailable
  • Major technical challenges
  • Requirement changes
  • Substantial rework needed

Ratio Guidelines

Typical Ratios

| Complexity | O:M:P Ratio | Example | |————|————-|———| | Simple tasks | 1:1.5:2 | 2:3:4 days | | Average tasks | 1:2:3 | 2:4:6 days | | Complex tasks | 1:2:4 | 2:4:8 days | | High uncertainty | 1:3:6 | 1:3:6 weeks |

Warning Signs

  • P/O > 10: Task too uncertain, break down further
  • M closer to O: May be overly optimistic
  • M closer to P: May be overly pessimistic

Aggregating Estimates

For Multiple Tasks 

Total Expected = Sum of Individual Expected Values
Total Variance = Sum of Individual Variances
Total Standard Deviation = √(Total Variance)

Project Level Example

| Task | O | M | P | Expected | Std Dev | |——|—|—|—|———-|———| | Analysis | 5 | 8 | 15 | 8.7 | 1.67 | | Design | 10 | 15 | 25 | 15.8 | 2.5 | | Development | 20 | 35 | 60 | 36.7 | 6.67 | | Testing | 8 | 12 | 20 | 12.7 | 2.0 | | Total | 43 | 70 | 120 | 73.9 | 7.35 |

Best Practices

Estimation Quality

  • Use historical data to calibrate estimates
  • Have multiple people estimate independently
  • Validate assumptions behind each estimate
  • Document reasoning for future reference

Team Involvement

  • Include subject matter experts
  • Consider different perspectives
  • Use Delphi technique for consensus
  • Review estimates regularly

Risk Considerations

  • Identify key assumptions
  • Plan for pessimistic scenarios
  • Build contingency based on uncertainty
  • Monitor actuals vs estimates

Common Pitfalls

Estimation Errors

  • Making all three estimates too similar
  • Anchoring on the first estimate given
  • Not considering external dependencies
  • Ignoring team experience differences

Process Mistakes

  • Using arithmetic mean instead of PERT
  • Not updating estimates as uncertainty reduces
  • Treating expected value as commitment
  • Ignoring confidence intervals

Tools and Templates

Simple Spreadsheet Formula 

=ROUND((B2+4*C2+D2)/6,1)  // PERT expected value
=ROUND((D2-B2)/6,1)       // Standard deviation

Monte Carlo Simulation

For complex projects, consider Monte Carlo simulation:

  1. Define probability distributions for each task
  2. Run thousands of simulations
  3. Analyze results for confidence levels
  4. Create realistic project schedules

Integration with Agile

  • Apply to user stories or epics
  • Use for release planning
  • Update estimates after each sprint
  • Track estimation accuracy over time