Critical Thinking Prompt Sheet

Put your draft through a fixed set of hard questions, aimed at the blind spots self-review tends to miss, and end with the three fixes that matter most.

Practice, not research: the question set itself is untested. The problem it counters is strongly supported: people check their own work for confirmation rather than refutation (Wason, 1960) and anchor on first impressions (Tversky & Kahneman, 1974). SMART is a weak, contested acronym (Doran, 1981), used here only as a completeness check.

Use it when:

• A report, proposal or recommendation is almost done and goes to someone who will decide or act on it.
• You wrote it yourself and have read it too many times to see its flaws.
• You review a colleague's draft and want questions that are specific, fair and hard to argue with.

Skip it when:

• The issue is whether the frameworks in the draft do any work: use 2.14 Framework Application Checklist.
• You suspect the sources are weak, not the reasoning: start with 2.15 Source Quality Ladder.

This set of questions is working craft without a canonical source: no study validates this particular list. What is well supported is the problem it counters. Self-review fails in predictable ways. People test their own ideas by looking for confirmation rather than for what would prove them wrong (Wason, 1960). They anchor on first framings and give too much weight to whatever comes easily to mind (Tversky & Kahneman, 1974). A fixed list of questions swaps good intentions for a procedure that works under time pressure too.

Two parts carry their own labels. The SMART test on recommendations is a practitioner acronym without theoretical or empirical grounding of its own (Doran, 1981), so use it to spot missing parts, nothing more. The decision hygiene methods in the optional section are consulting practice built on the bias research, not separately proven (Koller, 2024; Kahneman, Sibony & Sunstein, 2021). The questions that matter most aim at your blind side: the strongest objection, not a weak one, and the claims you cannot back up. They work best when someone else answers them about your draft too.

Compare what was asked with what you delivered. Then write the argument of the whole piece in one sentence and list any section that does not serve it.

Strong answer: "Asked: should we keep the small-cafe route? Delivered: a history of the route and three options, but no recommendation. Fix: add a recommendation on page 1." Main argument: "Keep the route, but raise the minimum order, because cafes leave over order size, not price."

What was asked for (the decision or brief):

What I delivered, and the gap:

My main argument in one sentence:

Sections that drift from that argument:

Illustrative example: the business, the roles and all numbers are made up.

The sales lead at the Gothenburg bakery from 2.1 has a draft report for the April board meeting. The head baker, who did not write it, runs the sheet on it too.

Step 1: The owner asked "keep, change or drop the cafe route?" The draft ended with three options and no choice. Main argument after the fix: "Keep the route and raise the minimum order, because cafes leave over order size, not price."

Step 2: The sales lead listed one claim without support. The head baker found a second: "the drivers can absorb the extra stops". Strongest objection: a higher minimum will drive away the cafes that still order.

Step 3: The recommendation lacked a start date, and the owner, not the sales lead, has the authority to accept losing customers.

Step 4: The two-page market history failed the so-what test and was cut to two lines.

Step 5: Priority fixes: add a date and owner, answer the objection with the interview data, cut the history. The two reviewers' red flags overlapped on only one item, which is why the second reader was worth it.