Problem Solving & Quality · State

Is / Is-Not Analysis

For each dimension — what, where, when, how much — write where the problem is and where it could be but is not. The contrast points at the cause.

Time45 min
FormatTeam
StageState

Is / Is-Not Analysis: what it is and why it works

Is / Is-Not analysis is a structured specification of a problem that records, for each dimension, where the problem is observed and the closest place where it could logically occur but does not. The four dimensions (What, Where, When and How much) turn a vague complaint into a bounded search area. The core logic is contrast: a true cause must explain why the defect appears on the IS side and why it is absent on the IS NOT side. Once both columns are filled, the team looks for what is distinctive about the IS side and what changed there, and those distinctions and changes become the candidate causes.

The method works because it replaces open-ended brainstorming with elimination. A fishbone can list forty plausible causes; an Is / Is-Not table often rules out most of them in minutes, because a cause that would also hit the IS NOT line cannot be the answer. It beats brainstorming when the problem is specific, fairly recent and has a clean comparison case: one line but not its twin, one product grade but not another. It is weaker for chronic problems with no sharp contrast. It sits naturally after a deviation statement and 5W2H, and it feeds directly into change analysis, which examines the distinctions and changes the table has surfaced.

What you need

A written deviation statement or first problem description
Production, quality and maintenance records for the period in question
People who know the process and the comparison cases: operators, technicians, quality
Samples or data from both affected and unaffected units where possible

What you get

A completed four-row IS / IS NOT table backed by facts
A list of distinctions between the IS and IS NOT sides
A dated list of changes linked to those distinctions
A short list of candidate causes that explain both columns, ready for verification
Data-collection actions for any cell that could not be filled

When to use it

When everybody describes the problem differently and the search area keeps growing.

How to do it, step by step

Draw four rows — What, Where, When, How much — and two columns: IS and IS NOT.
Fill the IS column with facts only: which product, which line, which shift, how many units, what trend.
For each IS, name the closest case where the problem could logically be but is not.
Write what is distinctive about the IS side compared with the IS NOT side, then what changed there.
Test each candidate cause: it must explain both columns. Keep only those that do.

Worked example: Porosity on one of two die-casting cells

Illustrative scenario — figures are realistic but not from a real company.

A mid-size aluminum die-casting plant supplying automotive housings sees X-ray porosity rejects on a transmission cover jump from 1.5% to 6%. Two identical 800-ton cells run the part. Early theories in the morning meeting blamed the alloy supplier, summer humidity and a newly hired operator.

What: porosity IS on the transmission cover, concentrated at the thick boss near the gate; it IS NOT on the pump housing cast from the same alloy lot, nor on the thin walls of the same part.
Where: it IS on cell 4; it IS NOT on cell 5, which runs the same die design, the same alloy lot and the same shift crews.
When: it IS present since the week of May 12, worst in the first hour after each start-up; it IS NOT present before that date, nor later in the shift.
How much: about 6% of castings, always at the same location on the part; the defect IS NOT scattered randomly.
Distinctions and changes: only cell 4 had its die thermal control unit replaced on May 10, after a breakdown, and the replacement unit had a lower heating capacity. Alloy, humidity and the new operator were dropped because they applied equally to cell 5.

Result. Die-surface temperature logs showed cell 4 reaching setpoint about 40 minutes later than cell 5 after start-up. Installing a correctly sized thermal unit and adding a warm-up criterion before releasing parts brought rejects back to 1.4% within two weeks. The team noted that the IS NOT column had eliminated three popular theories in less than an hour.

Open it in the FieldKit Problem Solving app → Get the Problem Solving template pack — join the waitlist

Common pitfalls and how to avoid them

Filling the IS NOT column with 'everything else'.Name the closest comparable case (twin line, sister product, previous week) so the contrast is sharp enough to eliminate causes.
Writing suspected causes or opinions into the IS column.Enter only facts with a source; park hypotheses on a separate list until the table is complete.
Skipping the How much row.Record quantity, proportion, trend and location on the part; extent often separates a step change from a gradual drift.
Keeping a candidate cause that explains only the IS side.Test every candidate against each IS NOT entry; if it would also produce the problem there, drop it or state the extra assumption it needs and check that assumption.

Frequently asked questions

What is the difference between Is / Is-Not and 5W2H?

5W2H collects the basic facts about a problem by answering seven open questions. Is / Is-Not goes a step further: for each dimension it also records where the problem could be but is not, and that contrast is what lets you eliminate causes. In practice, 5W2H is a good first pass on a vague report, and Is / Is-Not is the tool to use once you have enough facts to start narrowing down the cause.

Can Is / Is-Not analysis be used for chronic problems?

It can, but it is less powerful. The method relies on sharp contrasts, which are easiest to find when a problem starts suddenly or affects only part of a population. For a problem that has always existed everywhere, look for contrast in extent instead: which weeks, grades or machines are worst and which are best. Stratification of historical data usually has to come first.

How do you use Is / Is-Not to test possible causes?

Take each candidate cause and ask whether it explains every IS entry and every IS NOT entry. If the suspected cause is a new resin lot, but that lot also runs on the unaffected line, the candidate fails unless an extra assumption explains the difference. Candidates that pass are then verified with data or a controlled trial; the table alone never proves a cause.

Origin

Is / Is-Not specification — Charles H. Kepner & Benjamin B. Tregoe, The Rational Manager, 1965.

Used in these playbooks

Quality alert: the first 24 hours 1 day

One day to take control of a fresh incident: make safe, protect the customer, block every suspect lot, keep the evidence intact and pin down where the problem is — and is not.

Related methods

More in “State”

Describe the problem as a measurable gap — no cause, no culprit, no solution yet.