Review condition combinations with a pair-coverage register
A review can exercise every individual option while missing combinations that behave differently. Make coverage inspectable before calling a fixture set broad. This fictional practice exercise uses three binary factors in an invented document preview. It illustrates a coverage method; it does not describe Kiguri controls, executed tests, or evidence that a product is correct.
Define the factors and the coverage question
Call the factors X, Y, and Z. X is layout: compact is 0 and full is 1. Y is notes: hidden is 0 and shown is 1. Z is orientation: portrait is 0 and landscape is 1. The toy contract allows all eight three-factor combinations.
The question is whether every pair of factor values appears in at least one proposed fixture. For each factor pair, the possible value pairs are 00, 01, 10, and 11. Record these twelve coverage entries separately: four for XY, four for XZ, and four for YZ. Testing both values of X somewhere does not establish that X has appeared with every value of Y.
Map four fixtures to their pair entries
Choose the proposed fixtures 000, 011, 101, and 110. Keep the factor order X, Y, Z fixed so the digits remain meaningful. Fixture 000 covers XY=00, XZ=00, and YZ=00. Fixture 011 covers XY=01, XZ=01, and YZ=11.
Fixture 101 covers XY=10, XZ=11, and YZ=01. Fixture 110 covers XY=11, XZ=10, and YZ=10. Put these mappings in a shared coverage register, using one entry for each fixture and explicit references to the pair entries it covers. Reviewers can then check the map without interpreting a general statement that all combinations were considered.
Verify each pair independently
Read the XY entries across the four fixtures: 00, 01, 10, and 11. Read XZ: 00, 01, 11, and 10. Read YZ: 00, 11, 01, and 10. Each pair therefore contains all four possible value combinations in this specified fixture set.
Ask another reviewer to reconstruct one mapping from the digits alone. If factor order changes halfway through the record, an apparently complete register may be wrong. Also check that no fixture was accidentally duplicated or substituted. Coverage is a property of the actual selected conditions, not of their number or the confidence of the reviewer.
Keep the untested triples explicit
Four complete combinations remain outside the selected set: 001, 010, 100, and 111. A problem that occurs only when layout is full, notes are shown, and orientation is landscape would require 111; it is not established by the pair coverage above.
Do not rename this result exhaustive testing. Pair coverage answers the bounded question about two-factor combinations, while higher-order interactions remain unchecked. If a real requirement makes a combination infeasible, define that restriction first and rebuild the coverage question. Quietly deleting an impossible fixture can remove the only occurrence of a required feasible pair.
Hand over coverage and uncertainty together
The finished artifact includes factor definitions, feasibility assumptions, selected fixtures, pair mappings, and unchecked triples. A separate test process would still need expected results and actual observations; this selection exercise supplies neither a passing run nor release approval.
Ask the review owner whether the remaining combinations need examination for the actual change under consideration. Preserve that decision and its limits. A useful fixture plan makes its coverage reproducible and its gaps easy to find, rather than using a small orderly set as a promise that every possible interaction has been checked.
