While preparing the public campaign for Second-Pass Audit, I had an article, social copy, an email, metadata, links, and a distribution plan. The package looked nearly finished. Comparing it with the governing sources still found outdated positioning, the wrong call to action, incorrect public decision labels, an outdated URL, and a proof claim stronger than the record supported.
The review changed the decision from SHIP to HOLD while those issues were addressed. Better prose alone would not have fixed it. The offer, instructions, and evidence had to agree across the package.
That is why I separate generation from acceptance. AI can produce work worth considering. Someone still has to establish whether it is ready for the decision ahead: publishing a campaign, presenting a recommendation, signing a proposal, or acting on advice.
Key Takeaways
- Fluent AI output does not establish that a deliverable is ready to use.
- Compare work with its governing source, correct material differences, and recheck affected items.
- SHIP, REVISE, or HOLD should reflect the evidence, not the draft’s polish.
Why “What did you miss?” needed a source
The habit began with a question I kept asking ChatGPT after important answers: “What did you miss?” Sometimes the second answer surfaced a useful omission. Sometimes it produced another interpretation that sounded just as confident as the first.
So I changed the assignment. Return to the material that should govern the answer and compare the two directly. For a campaign, that might mean the approved brief, current offer, and evidence for its claims. For a proposal, it might mean the agreed scope, exclusions, and terms.
The source gives the review something testable. “Make this stronger” can invite another rewrite. “Does this match the approved offer?” asks a question the record may actually answer.
What the campaign review established
The record shows an acceptance decision changing because the comparison found material discrepancies. It does not measure error reduction or business outcomes across repeated use.
The practical lesson is still useful. A correction has to reach the places where the problem appears. Fixing an article’s offer language leaves the package inconsistent if the email still makes the old promise. Updating the main link leaves a broken path if a secondary link retains the retired destination.
The recheck covers what changed and what depends on it. Parts already supported should be preserved.
Compare, correct, recheck, then decide
Provide three things: the work you are considering accepting, the source that should govern it, and what you intend to do with it. The operating sequence is COMPARE → IDENTIFY → CORRECT → RECHECK → DECIDE.
- Compare. Inspect the work directly against the relevant evidence, instructions, records, requirements, or terms.
- Identify. Find material omissions, errors, unsupported conclusions, weak interpretations, underweighted details, and missed constraints. A finding matters when it could change the intended decision or use.
- Correct. Apply the correction the source supports. Do not rewrite sound material simply to produce a different answer.
- Recheck. Inspect the affected claims, calculations, recommendations, links, metadata, or dependent work.
- Decide. Judge the final state against the intended use.
The current method uses three acceptance outcomes:
- SHIP: No unresolved material issue remains for the intended use.
- REVISE: The available source establishes a material correction that has not yet been applied or rechecked.
- HOLD: Missing evidence, unresolved authority, conflicting sources, or required qualified review prevents responsible acceptance.
If the brief supplies the correct URL, fix the wrong link and recheck it. If required production approval is pending, another edit cannot supply that authority. The earlier campaign review recorded HOLD; the guidance above makes the current boundary more explicit.
Once corrections and affected checks pass, the final decision can become SHIP. That means ready for the stated use. It does not itself grant permission to publish, send, spend, or launch. And when the work already meets the requirements, the audit should say so and stop.
Seven decisions. One acceptance standard.
AI can make work look finished. Can you decide when it deserves approval? Compare the source against the AI output and choose SHIP, REVISE, or HOLD in seven short challenges.
What is the appropriate decision?
- Correct / next
- Recheck
Your seven decisions
What matters outside this exercise: Return to the actual source, correct material differences, recheck affected work, and then decide. A simulated challenge cannot verify a real proposal or replace required expert approval.
Teaching scenarios are fictional. SHIP, REVISE, and HOLD follow the article’s stated decision definitions. No personal information is requested. If site analytics are configured, anonymous interaction events may be counted.
Use the source that fits the decision
I have also used this discipline when reviewing advice, proposals, contracts, recommendations, and personal decisions. The source changes with the situation. A business recommendation may need customer evidence and financial assumptions; a contractor proposal may need the agreed scope, materials, exclusions, warranties, and competing bids.
Matching the source does not prove the source is right. An outdated document can produce a perfectly faithful, unusable recommendation. In medical, legal, financial, technical, engineering, or compliance decisions, the audit may help identify a question for a qualified professional rather than settle it.
The seven challenges above are fictional teaching scenarios. They let you inspect the distinctions; their score does not validate a real proposal or qualify someone to make a high-stakes decision.
Try it on work you are about to use
Choose one deliverable, provide its governing source, and state the intended decision. Use the prompt below. If the AI cannot access the source, it should identify that limit rather than pretend to have compared it.
Run a Second-Pass Audit. Work or recommendation: [paste or attach what I am considering accepting] Decision-relevant source: [paste, attach, or link the material that should govern it] Intended use / decision: [state what I am about to approve, present, publish, sign, buy, follow, or act on] Compare the work directly against the source before acceptance. Check for material omissions, factual or interpretive errors, unsupported conclusions, weak interpretations, underweighted details, missed constraints or qualifiers, changed emphasis, and downstream inconsistencies. Preserve supported content. Do not rewrite for novelty. Return: 1. Decision: SHIP, REVISE, or HOLD. 2. Short rationale. 3. Material issues, each with its source basis and required correction. 4. Affected downstream elements that need rechecking. 5. Next action. Matching the source does not prove the source is right. If the source is insufficient or consequential expert judgment remains unresolved, return HOLD.
Read the findings against the source before accepting them. If correction is authorized, apply supported changes and recheck what they affect. Get missing evidence or required approval when the comparison cannot resolve the issue. Use a visual explanation only when it makes a dependency, gap, or consequence easier to understand; it must carry the same findings and decision as the written audit.
Resources and the rest of the series
You can start with the prompt above in the AI tool you already use. For the supporting resource, download the Complete Second-Pass Audit Kit. If you want to inspect the reusable instructions and source files, view the public GitHub repository. GitHub is optional; you do not need it to use the method.
- Part 2: What Did You Miss? How a Simple Question Became My Second-Pass Audit
- Part 3: Before You Accept the Answer: Using a Second Pass for Work, Health, Wealth, and Real-World Advice
