Is / Is Not analysis describes a problem along four dimensions — what, where, when and extent — and for each one records what is affected and what could reasonably have been affected but is not. That boundary is the evidence: a candidate cause that cannot explain both columns is not the cause. It is the standard tool for 8D step D2, and it works just as well outside a factory.

Download the Is / Is Not template

Ten pre-filled questions across the four dimensions, with columns for IS, IS NOT, distinction and change. Fill the first two columns from evidence, then work left to right. No signup, no email.

📈
Free

Is / Is Not — Excel (.xlsx)

Questions pre-filled, panes frozen so the dimension stays visible as you scroll right.

Download .xlsx
📝
Free

Is / Is Not — Word (.docx)

Landscape table for pasting into a D2 section of an 8D report or a customer-facing document.

Download .docx
📄
Free

Is / Is Not — PDF (print)

Printable landscape A4 — the version that works on a wall in front of the team.

Download .pdf

The four dimensions

The questions are deliberately narrow. Each one has a factual answer that somebody can check, which is what stops D2 from turning into a discussion about causes before anyone has agreed what the problem is.

DimensionQuestionsWhat it pins down
WhatWhich object, part or product? What is the defect or deviation?Separates the thing from the symptom — two defects on one part are two analyses.
WhereWhere geographically — site, line, cell? Where on the object does it appear?Location on the object is the question most often skipped, and it frequently points straight at a fixture or a station.
WhenFirst observed? When in the process or lifecycle? Continuous, intermittent, every nth?A start date to line changes up against, and a pattern that rules whole classes of cause in or out.
ExtentHow many objects? How many defects per object? Rising, flat or falling?Scale and direction — a falling trend after an unrecorded change is a clue in itself.

A worked example

Illustrative, not a real case: labels lifting at the leading edge on 500 ml bottles. The complaint arrives as “labels are coming off”, which is not something you can test a cause against.

QuestionISIS NOTDistinction
Which product?500 ml bottle, product A250 ml bottle, run on the same lineLarger label, curved surface
Which defect?Label lifts at the leading edgePrint quality, position, wrinklingAdhesion only, at one edge
Which line?Line 2Lines 1 and 3, same productLine 2 has the older applicator
Which shift?Day shiftNight shift, same lineDay shift runs at higher throughput
Since when?4 AugustNot before 4 AugustSomething changed on or before that date
What pattern?Intermittent, roughly one pallet in eightNot continuousBatch-related rather than machine-constant
How many?~12% of affected pallets, trend risingNot all bottles on affected palletsSomething varying within a run

Now the two questions that do the work. First: what is different about the IS? Line 2, day shift, larger curved label, since 4 August, intermittent. Second: what changed in or around those differences? An adhesive lot loaded on line 2 on 3 August, and nowhere else.

That candidate has to survive the whole grid: it explains line 2 and not lines 1 and 3 (lot loaded only there), it explains the start date, and it explains intermittency if the lot is mixed with the previous one in the reservoir. It does not obviously explain the day/night difference — so either throughput interacts with the adhesive’s open time, or the shift distinction is a coincidence worth checking. That unresolved cell is not a nuisance; it is the next test.

Distinction, then change

The step teams miss is the second one. A distinction is anything that separates the IS from the IS NOT — but most distinctions have been true for years and cannot explain a problem that started last week. The useful question is what changed in or around a distinction, and when.

PatternWhat it suggests
Distinction is old, change is recent and lines up with the start dateStrongest lead. Test this first.
Distinction is old, no change foundEither the change is unrecorded, or your start date is wrong. Both are worth verifying before hunting further.
Change found but it predates the problem by monthsProbably not the trigger on its own — look for what made a dormant condition matter now.
No distinction at all between IS and IS NOTThe IS NOT column is not specific enough yet. Add near misses until something separates.

Four ways the grid goes wrong

The IS NOT column lists impossibilities. “Not our competitor’s product” is true and useless. The column only informs when it holds near misses — the variant that runs on the same line, the shift that uses the same machine, the region on the same release.

Causes get written into the description. “Where: line 2, because the applicator is worn” ends the analysis before D4 starts. D2 is facts only; the applicator is a hypothesis to be tested against the grid.

The grid is filled from memory. Every cell should be traceable to a record, a measurement or an inspection someone actually did. A grid built from what people remember produces confident boundaries that are not real.

One contradiction is ignored. A candidate cause that explains six rows and contradicts the seventh is not 85% right — it is unproven. The contradicting row is usually where the real mechanism is hiding, as with the shift difference in the example above.

Where this sits in 8D — and what comes next

Is / Is Not belongs in D2, describe the problem, between assembling the team and putting containment in place. Its output is a problem statement precise enough that D3 containment can be scoped and D4 root cause has something to test against. Teams that skip D2 tend to arrive at D4 with several plausible causes and no way to eliminate any of them.

Next stepToolWhy it follows
Record the whole discipline sequence8D report templateD0–D8 in one document; the grid drops into D2
Write the statement itselfProblem statement generatorFlags blame and hidden solutions as you type
Follow the chain from the surviving candidate5 WhysThe grid narrows to one candidate; 5 Whys goes down from it
Nothing survives the gridFishboneGenerates more candidates across categories, then retest them here
Several defect types competing for attentionPareto analysisDecide which problem deserves a grid at all

FAQ

What is Is / Is Not analysis?

A problem specification technique that describes a problem along four dimensions — what, where, when and extent — and for each records both what is affected and what could reasonably have been affected but is not. The boundary between the two is the evidence: any candidate cause has to explain every IS and every IS NOT.

What are the four dimensions of an Is / Is Not analysis?

What (which object, which defect), where (which location, and where on the object), when (first observed, when in the process, what pattern), and extent (how many objects, how many defects each, and the trend). Together they turn a vague complaint into a specification precise enough to test causes against.

What goes in the IS NOT column?

Only things that could plausibly have been affected but are not — the other shift, the other line, the 250 ml variant, the region on the same release. Listing something with no reasonable chance of being affected adds nothing; the value comes from near misses, because each one narrows where the cause can live.

How does Is / Is Not lead to a root cause?

Two steps. What is different about the IS compared with the IS NOT — a distinction. Then what changed in or around that distinction, and when. Causes hide in changes to differences: a distinction that has been true for years and a change that lines up with the start date are very different leads.

Where does Is / Is Not fit in the 8D process?

In D2, describe the problem — the step that turns a complaint into a fact-based statement before D3 containment and D4 root cause. Teams that skip it usually reach D4 with a problem defined loosely enough that several causes look plausible and none can be ruled out.

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

5W2H asks who, what, where, when, why, how and how much, to make sure nothing is missing from the description. Is / Is Not takes several of the same dimensions and adds the negative case — what is equally exposed but unaffected. 5W2H gives you a complete description; Is / Is Not gives you a boundary to test candidate causes against.

Can Is / Is Not be used outside manufacturing?

Yes. The dimensions translate directly to software and service work: which endpoint and which error, which region or tenant, when it started and whether it is intermittent, how many requests and whether the rate is rising. One region failing and the others fine is the same analytical situation as one line producing defects and the others not.

Related