BETAPlatform actively being built — new topics and features added regularly.

ISTQB Foundation Level (CTFL 4.0.1)~5 min read19/26

Test Estimation & Prioritization

// techniques for estimating test effort and prioritizing test cases.

loading...
// content

Unrealistic test estimates cause project failures — either by over-promising or by being ignored

A tester who estimates "two weeks for testing" without a basis for that number will either be held to it when it is wrong, or dismissed as someone who does not understand the project. Either outcome damages trust and quality.

Test estimation is the process of approximating the effort, time, and resources needed for testing activities. Good estimates are transparent, justified, and continuously revised as more information becomes available.

// example: netflix — estimating a major platform overhaul

Scenario: Netflix's QA team is asked to estimate testing effort for a redesign of the content recommendation engine — a system that affects every user's home screen globally. What happened: The team used a combination of expert estimation (senior testers with experience of similar overhauls) and metrics-based estimation (historical data: the previous recommendation engine update took 340 test-hours for 120 test cases). They identified that the new engine had 180 test conditions — scaling proportionally to 510 test-hours, adjusted upward by 20% for the additional complexity of personalisation algorithms. The final estimate was 612 test-hours, communicated as a range (580–650) to reflect uncertainty. Why it matters: Presenting a range with justification rather than a single number builds credibility and helps stakeholders understand the basis for the estimate.

Test Estimation Techniques — CTFL 4.0.1

Expert-based estimation

Relies on the experience and judgement of skilled testers or managers. Techniques include:

  • Wideband Delphi — multiple experts estimate independently, discuss discrepancies, and converge on a consensus
  • Planning poker — common in Agile; team members select estimation cards simultaneously to avoid anchoring bias

Metrics-based estimation

Uses historical data from previous projects to derive estimates. Approaches include:

  • Extrapolation — if previous similar feature took X hours per test case, estimate new feature by scaling
  • Test point analysis — assigns complexity weights to test conditions and converts to effort

Three-point estimation

Produces a more realistic estimate by combining three scenarios:

  • O (Optimistic) — best case if everything goes smoothly
  • M (Most likely) — the expected case based on experience
  • P (Pessimistic) — worst case with realistic problems

Formula (PERT): Expected Estimate = (O + 4M + P) ÷ 6

// tip: Exam Tip: Know the three-point estimation formula: (Optimistic + 4 × Most Likely + Pessimistic) ÷ 6. The exam may give you three values and ask for the PERT estimate. Also remember that estimates should always be presented as ranges with stated assumptions — not as single definitive numbers.

Three-Point Estimation — Worked Example

Scenario: Estimate the testing effort for a new user authentication module.

ScenarioAssumptionEstimate
Optimistic (O)Requirements are clear, no environment issues, 1 experienced tester40 hours
Most Likely (M)Some requirement clarifications needed, minor environment delays70 hours
Pessimistic (P)Requirements change mid-testing, environment instability, defect rate is high130 hours

PERT Calculation:

Expected = (O + 4M + P) ÷ 6 = (40 + 4 × 70 + 130) ÷ 6 = (40 + 280 + 130) ÷ 6 = 450 ÷ 6 = 75 hours

Present as a range: 65–85 hours, with stated assumptions. The PERT estimate weights the most likely scenario four times heavier than the extremes, producing a more realistic result than a simple average.

Three-Point (PERT)

Basis

Optimistic, Most Likely, Pessimistic scenarios

Best for

High-uncertainty tasks; when range is more useful than point estimate

// Pros

  • Accounts for uncertainty

  • Produces realistic range

  • Weights most likely scenario heavily

// Cons

  • Requires honest pessimistic estimation

  • Still subjective

  • Formula may feel arbitrary to stakeholders

// PERT Calculator

Formula: (O + 4M + P) ÷ 6

Optimistic (O)

Most Likely (M)

Pessimistic (P)

// Result

75 hours

Present as range: 6585 hours

// Example

O=40h, M=70h, P=130h → (40 + 4×70 + 130) ÷ 6 = 75h (present as 65–85h range)

// Exam tip

Estimates should ALWAYS be presented as ranges with stated assumptions — not single numbers. The exam will test whether you understand that estimates are approximations, not commitments.

Test Estimation Techniques Compared

TechniqueBasisBest Used WhenLimitation
Expert judgementTester experience and intuitionEarly in project with little data; experienced teamSubjective; anchoring bias risk
Wideband DelphiMultiple expert consensusComplex or unfamiliar systems; multiple stakeholdersTime-consuming; requires expert availability
Planning pokerTeam consensus with cardsAgile sprints; team has shared contextOnly works with shared understanding of scope
Metrics-basedHistorical project dataRepeat project type; historical data existsRequires comparable historical data
Three-point (PERT)Optimistic, likely, pessimistic scenariosHigh-uncertainty tasks; when range is more useful than a pointRequires honest pessimistic estimation

// warning: Exam Trap: "An estimate is a commitment." This is false. An estimate is an approximation based on current knowledge and stated assumptions. As the project progresses and more information becomes available, estimates should be revised. Presenting an estimate as a fixed commitment without stating assumptions creates unrealistic expectations and damages trust when reality differs.

Exam Practice Questions

// ctfl 4.0.1 style — select an answer to reveal explanation

3Q
Q1.A tester estimates a feature will take 20 hours (optimistic), 35 hours (most likely), and 60 hours (pessimistic). What is the PERT estimate?
Q2.Which estimation technique involves multiple experts estimating independently and then converging through structured discussion?
Q3.A test manager uses data from three previous similar projects to estimate effort for a new project. Which technique is being used?
// end