Using PERT to Give Estimates When You're Uncertain

When estimating software projects, there are two scenarios:

When you're certain: You can confidently commit to a date. "Yes, I can deliver this by Friday."

When you're uncertain: Be honest about it. Don't guess a single date and hope for the best. Use PERT instead.

What is PERT?

PERT (Program Evaluation and Review Technique) is an estimation method that acknowledges uncertainty by using three estimates instead of one:

• Best case (Optimistic): Everything goes smoothly, no surprises
• Normal case (Most Likely): Realistic scenario with typical challenges
• Worst case (Pessimistic): Murphy's Law kicks in, unexpected issues arise

The PERT Formulas

PERT calculates a weighted average that accounts for uncertainty:

Expected = (Best + 4 × Normal + Worst) / 6
σ = (Worst - Best) / 6

The formula weights the most likely estimate more heavily (4×) since that's where reality usually lands. The standard deviation (σ) captures how uncertain each estimate is.

Example: Estimating a Feature Set

Feature	Best	Normal	Worst	Expected	σ
Feature 1	1	3	6	3.2 days	0.83
Feature 2	2	4	8	4.3 days	1.00

If Sequential:

• Total: 7.5 days
• Combined σ: 1.3 days
• 95% range: ~5-10 days

Instead of committing to a single number, you communicate a range with confidence levels.

Why This Works

1. Avoid domino effect: Communicating uncertainty prevents downstream teams from building plans on shaky commitments
2. Sets realistic expectations: A range prepares everyone for variability
3. Reduces stress: You're not locked into a single number you might miss
4. Encourages discussion: The range opens conversation about priorities and trade-offs

When to Use PERT

• New technology or unfamiliar codebase
• Complex integrations with external systems
• Features with unclear requirements
• Projects with dependencies on other teams

When you're uncertain, don't pretend to be certain. Use PERT and communicate the range.