COMPACT PRODUCT TEARDOWN WORKSHEET
Symptom → user cost → possible cause → smallest test

Original worksheet by Alfred.

PURPOSE
Turn a product complaint or observed friction into a bounded investigation without
presenting an inferred cause, preferred fix, or wider impact as fact.

This worksheet is a reasoning aid. A completed copy does not prove root cause,
prevalence, user impact beyond the observed scope, test completion, publication,
or a product result.

SAFE USE
- Use only evidence you are authorized to review and share.
- Remove names, contact details, account values, order identifiers, and private data.
- Prefer synthetic data for reproduction when it can answer the question.
- Keep restricted evidence restricted; publish only the minimum safe aggregate.
- Do not identify, embarrass, blame, or speculate about an individual.
- Do not expose a live exploit, credential, sensitive route, or unsafe procedure.
- Stop when consent, authority, rights, privacy, or safety is uncertain.

EVIDENCE STATES
Use one state for every review lane. Never turn a blank into PASS.

PASS
  The declared check ran within scope and its acceptance condition was met.

FAIL
  The declared check ran within scope and its acceptance condition was not met.

BLOCKED
  A required check could not run because an authority, access, dependency, or
  safety prerequisite was unavailable.

NOT_TESTED
  No eligible check ran.

SUPERSEDED
  The evidence no longer binds the current artifact, environment, hypothesis,
  or test revision.

UNCERTAIN
  Evidence exists, but its identity, scope, quality, or interpretation cannot
  support a narrower state.

PART A — RECORD IDENTITY AND SCOPE

Record ID:
Record revision:
Created at, with timezone:
Reviewed at, with timezone:
Reviewer role, if needed:

Product or flow label (non-identifying):
Artifact or build identity:
Environment and version:
Task being attempted:
Observed step:
Evidence source type:
Evidence observed at, with timezone:
Authorized evidence scope:
Restricted evidence location or reference (do not copy private data here):

Known scope boundaries:
- Input method:
- Browser, device, or client:
- Locale or language:
- Account state, expressed generically:
- Network or dependency state:
- Number of eligible observations:
- Time interval represented:
- What was not observed:

Identity and scope state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

PART B — RIGHTS, PRIVACY, AND SAFETY GATE

[ ] Authority to inspect the evidence is confirmed.
[ ] Authority to run the proposed test is confirmed.
[ ] Evidence needed for the decision is separated from unrelated data.
[ ] Names, contact details, account data, payment data, and identifiers are removed.
[ ] Any retained example uses synthetic or properly consented data.
[ ] Public wording will describe a product pattern, not target an individual.
[ ] No copied media or material without a known rights basis is included.
[ ] No credential, secret, sensitive route, exploit detail, or unsafe procedure appears.
[ ] Stopping conditions are explicit.
[ ] Uncertainty will remain visible.

Gate state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Evidence or blocker:

STOP RULE
If this gate is not PASS, do not run or publish the teardown. Preserve the narrow
record privately only when authorized.

PART C — SYMPTOM: WHAT HAPPENED?

Write an observable event. Do not infer intent, competence, cause, or prevalence.

Task start condition:
Action taken:
Expected visible or audible result, if specified:
Observed visible or audible result:
Step where progress changed:
State before the event:
State after the event:
What did not happen:
What can be reproduced directly:
What is reported but not directly observed:
What remains unknown:

One-sentence bounded symptom:

Avoided interpretation or motive claim:

Evidence supporting the symptom:
1.
2.
3.

Evidence contradicting or narrowing the symptom:
1.
2.
3.

Symptom state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

PART D — USER COST: WHICH OUTCOME WAS BLOCKED OR HARDER?

Observed task outcome:
Observed extra work, delay, confusion, or recovery step:
Observed loss of entered work, if any:
Observed accessibility barrier, if any:
Observed uncertainty about completion, if any:

Bounded observed cost:

Potential wider cost to investigate, clearly labeled as a hypothesis:

Claims not supported by this evidence:
[ ] Population prevalence is unknown.
[ ] Revenue effect is unknown.
[ ] Conversion effect is unknown.
[ ] Customer-retention effect is unknown.
[ ] Causality beyond the observed scope is unknown.
[ ] Other:

User-cost state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

PART E — POSSIBLE CAUSE: A FALSIFIABLE HYPOTHESIS

Do not write “root cause” unless evidence has actually established it.

Condition:
Possible mechanism:
Observed symptom it could produce:
Specific evidence expected if true:
Specific evidence expected if false or incomplete:

Hypothesis sentence:
When [condition], the product may [mechanism], causing [bounded symptom].
If this is true, we expect [specific evidence].

Supporting evidence:
1.
2.
3.

Contradicting evidence:
1.
2.
3.

Plausible alternative A:
Evidence that would distinguish it:

Plausible alternative B:
Evidence that would distinguish it:

Assumptions:
1.
2.
3.

Possible-cause state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

PART F — SMALLEST TEST: REDUCE THE LARGEST UNCERTAINTY

Largest decision-relevant uncertainty:
Decision this test can change:
Why a smaller or safer check cannot answer it:

Test identity:
Test revision:
Method:
Environment:
Synthetic, consented, or restricted data:
Exact condition changed:
Conditions held fixed:
Evidence to collect:
Evidence deliberately not collected:

Supports the hypothesis if:

Weakens the hypothesis if:

Leaves the result uncertain if:

Stopping conditions:
- Authority or consent changes:
- Private or identifying data appears:
- A safety or security concern appears:
- The environment no longer matches:
- The test could affect a real user or live transaction:
- Other:

Acceptance state vocabulary:
SUPPORTED_WITHIN_SCOPE
  The declared expected evidence appeared within the bounded test.

WEAKENED_WITHIN_SCOPE
  Declared contrary evidence appeared within the bounded test.

INCONCLUSIVE
  The test ran, but evidence cannot distinguish the named explanations.

BLOCKED
  The test could not run or complete under the declared gate.

SUPERSEDED
  The artifact, environment, hypothesis, or protocol changed.

STOP
  Rights, privacy, authority, safety, or quality made the test ineligible.

Planned test state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

PART G — RESULT RECORD
Complete only after the exact test has run. Otherwise keep every lane NOT_TESTED.

Test started at, with timezone:
Test ended at, with timezone:
Exact artifact or build tested:
Protocol revision used:
Environment actually observed:
Data type actually used:
Stopping condition triggered:

Observed evidence:
1.
2.
3.

Unexpected evidence:
1.
2.
3.

Missing evidence:
1.
2.
3.

Result: SUPPORTED_WITHIN_SCOPE / WEAKENED_WITHIN_SCOPE / INCONCLUSIVE /
        BLOCKED / SUPERSEDED / STOP / NOT_TESTED

Reason:

What remains unknown:

Next bounded decision:

Do not write:
- “fixed” when only a prototype or local path passed;
- “root cause” when alternatives remain plausible;
- “users” when evidence covers one bounded observation;
- “zero impact” when impact was not measured;
- “published” when the artifact is only local, staged, uploaded, or private.

PART H — PUBLIC-RELEASE REVIEW
Complete separately from the investigation. Passing a test is not permission to publish.

[ ] The symptom is observable and bounded.
[ ] The user cost separates observed evidence from possible wider impact.
[ ] The cause remains labeled as a hypothesis unless established.
[ ] Supporting and contradicting evidence are both preserved.
[ ] The smallest test names evidence that supports and weakens the hypothesis.
[ ] Test state and publication state are separate.
[ ] No prevalence, revenue, conversion, customer, or causal claim was invented.
[ ] No individual is identified, blamed, embarrassed, researched, or pressured.
[ ] No personal, account, payment, contact, or private operational data appears.
[ ] Rights to all included text and media are recorded.
[ ] No sensitive exploit, credential, route, or unsafe procedure appears.
[ ] The AI authorship disclosure is visible.
[ ] Sources support only the claims attributed to them.
[ ] Proposed, completed, reproduced, and publicly verified work are distinguished.
[ ] Contradictory and missing evidence remains visible.
[ ] Later evidence can supersede the diagnosis without erasing history.

Public-release state: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

PUBLICATION STATE
Choose exactly one:
[ ] LOCAL_DRAFT
[ ] ASSEMBLED_LOCAL_CANDIDATE
[ ] VALIDATED_LOCAL_CANDIDATE
[ ] QUEUED_FOR_AN_AUTHORIZED_DESTINATION
[ ] UPLOADED_NOT_PUBLICLY_VERIFIED
[ ] PUBLICLY_VERIFIED
[ ] SUPERSEDED
[ ] WITHDRAWN_AND_REMOTELY_VERIFIED

Destination, only if authorized:
Public route, only if verified:
Verified at, with timezone:
Verification evidence reference:

PART I — FAILURE-SHAPED REVIEW TESTS
Run these against the completed record. A failure is useful evidence.

TEST 1 — MOTIVE LEAK
Replace the symptom with a claim about a team’s intent or competence.
Expected: FAIL; rewrite as an observable event.
State:

TEST 2 — SCOPE LEAK
Use one observation to claim population-wide prevalence.
Expected: FAIL; preserve the observed scope and mark prevalence unknown.
State:

TEST 3 — COST INFLATION
Turn a blocked task into an unsupported revenue or conversion claim.
Expected: FAIL; separate observed task cost from business impact.
State:

TEST 4 — CAUSE AS FACT
Remove “possible” or “hypothesis” before alternatives are tested.
Expected: FAIL; restore falsifiable language.
State:

TEST 5 — CONFIRMATION-ONLY TEST
Define what supports the hypothesis but not what weakens it.
Expected: FAIL; add contrary evidence before running the test.
State:

TEST 6 — FIX-AS-TEST
Propose a production change when a smaller synthetic reproduction can answer the question.
Expected: FAIL; choose the least costly safe discriminator.
State:

TEST 7 — PRIVATE-DATA COPY
Paste raw account, payment, contact, or identifying evidence into the public card.
Expected: STOP; remove it and reassess whether publication is safe.
State:

TEST 8 — BLANK-AS-PASS
Leave a required lane empty and call the teardown complete.
Expected: FAIL; use NOT_TESTED, BLOCKED, or UNCERTAIN.
State:

TEST 9 — CHANGED BUILD
Run the protocol on a different artifact but retain the old result.
Expected: SUPERSEDED; bind a new result to the exact tested identity.
State:

TEST 10 — PUBLICATION COLLAPSE
Treat local validation, a private upload, or an upload response as public evidence.
Expected: FAIL; require authorized logged-out public verification.
State:

TEST 11 — ONE PASS PROVES ROOT CAUSE
Use one successful reproduction to erase plausible alternatives.
Expected: FAIL; narrow the result to the tested condition.
State:

TEST 12 — LATER EVIDENCE OVERWRITE
Delete the old conclusion when corrected evidence arrives.
Expected: FAIL; preserve history and mark the earlier decision SUPERSEDED.
State:

FINAL COMPACT CHECK
1. Is the record bound to an exact artifact, environment, and observation time?
2. Is the symptom observable rather than interpretive?
3. Does the symptom state what was not observed?
4. Is observed user cost separate from possible wider impact?
5. Are prevalence, revenue, conversion, and causality left unknown unless evidenced?
6. Is the possible cause written as a falsifiable hypothesis?
7. Are at least two plausible alternatives visible when relevant?
8. Is contradicting evidence retained?
9. Does the smallest test reduce the largest decision-relevant uncertainty?
10. Does the test predeclare both supporting and weakening evidence?
11. Are stopping conditions explicit?
12. Are synthetic or consented data used where possible?
13. Are evidence states explicit rather than inferred from blanks?
14. Is the result narrower than the tested artifact, condition, and interval?
15. Are test completion and publication state separate?
16. Is all public material rights-safe and free of private information?
17. Does public wording avoid targeting or embarrassing anyone?
18. Is authorship disclosed consistently as Alfred?
19. Can corrected evidence supersede the decision without erasing history?
20. Does the final record avoid claiming work or results that were not verified?

FINAL REVIEW STATE: PASS / FAIL / BLOCKED / NOT_TESTED / SUPERSEDED / UNCERTAIN
Reason:

RIGHTS AND PROVENANCE
This worksheet, its field structure, state vocabulary, review tests, and compact
checklist are original work by Alfred. It contains no third-party
media, customer material, real account data, copied product evidence, or claimed
product result. A filled copy inherits the rights, privacy, authority, and accuracy
requirements of the evidence entered into it.