Red-team sheet vs devil's advocate: test the objection

A devil’s advocate performs disagreement; a red-team sheet records the objection, the evidence that would settle it and a verdict, so the challenge outlasts the meeting. The role puts a concern into the room. The sheet keeps that concern open after the room empties, when someone can inspect the records and the plan can still change.

Consider a hypothetical founder about to sign a lease for a second fulfilment site. Orders have risen and the current site feels crowded. Asked to play devil’s advocate, a colleague says, “What if the rise came from a promotion?” The founder replies that demand looks durable. The meeting moves on. The question nobody answers is whether ordinary demand will exceed the current site’s capacity for long enough to justify a fixed lease.

No real company or order record is described here. The example shows how an objection can become a decision check instead of a line of dialogue everyone remembers differently.

What is the difference between the role and the sheet?

A devil’s advocate is asked to argue against a proposal. That can be valuable: a well-stated objection may expose a missing assumption, or evidence the decision owner has overlooked. The role, however, does not say what must happen once the objection has been voiced. A lively exchange can end without anyone checking the point. And an argument made because it was assigned is easy to dismiss for the same reason.

A red-team sheet is a record built around one consequential claim. It states the strongest supported objection, what observation would count against the claim, who will check it and by when, and what the result does to the plan. It can also hold a question that stays unresolved. Filling in boxes is not the check; someone still has to inspect the evidence.

The two can work together. Ask a colleague to attack the lease, then use the sheet to carry the best attack past the meeting. For a cheap, easily reversed choice, a short conversation may be enough. The more a decision costs to undo, the more useful a record that others can review.

What goes on the red-team sheet?

Chapter 4 of The Evidence-Aware Life sets out six moves for testing a belief of your own that matters to a decision. The sheet below adapts them to a single decision. It is a proposed working form, not a score for the quality of a plan.

Move	Write down
1. Pick the belief worth attacking.	One checkable claim, the action it supports, and the cost of an error.
2. Steel-man the other side.	The strongest credible opposing case and the support it already has, stated so that someone who holds it would recognise it.
3. Hunt the disconfirming evidence.	What would count against the claim, what currently supports it, where to look, who will check, and by when.
4. Surface the hidden assumptions.	The necessary condition whose failure would most change the decision, and how to inspect it.
5. Run a pre-mortem.	One plausible way the action could fail, plus an early signal that would make you investigate.
6. Decide: keep, modify or drop.	The verdict, the reason, what remains unchecked, and a trigger to reopen the decision.

Apply the same evidence standard to the claim and to the objection, and write down what would count against the claim before you open the records. The chapter’s rule is that an objection earns weight through its support, not through the discomfort it causes. An uncomfortable argument is not automatically sound. Nor does a reassuring story show how often a plan works. The case-study guide explains why one visible success cannot supply a success rate.

If you cannot find a credible objection to the central claim, do not manufacture one. Chapter 4 suggests examining a boundary or an implementation assumption instead. For example: does the claim hold for this customer group, at this cost, with this team’s capacity? An honest blank is better than a weak adversary invented to make the plan look strong.

Worked example: should the founder sign the lease?

The founder’s claim is that ordinary order volume will exceed the current site’s usable capacity during the proposed lease term. The action is to sign now. A real decision would need the lease terms, the firm’s cash position and a capacity measure. The records and results below are invented solely to demonstrate the sheet.

The strongest opposing case has two parts. The recent rise may be concentrated in promotion weeks. And the apparent space problem may be a packing queue caused by shift scheduling. Asking “what if?” establishes neither; both can be checked.

The operations lead takes the disconfirming check: separate promotional from ordinary orders in the order log, then compare ordinary demand with the site’s measured throughput. A shift-by-shift packing log tests the capacity assumption. The finance lead compares the fixed lease obligation with the cost and limits of temporary overflow capacity. Both checks are due before the signing deadline. If a record cannot be obtained in time, the gap is marked rather than filled with a reassuring estimate.

Now suppose the invented 12-week order log shows that four promotion weeks account for the rise, while the other eight stay near the earlier ordinary level. The packing log places the queue at one shift handover; it shows no shortage of floor area across the day. The hidden assumption, that more space is the only way to handle ordinary orders, has lost its support. This hypothetical result does not establish that demand will never grow. It shows that the present case for signing this lease is weak.

The pre-mortem imagines the second site open but underused while the lease payments continue. Its early signal: ordinary-week demand staying below the current site’s measured capacity once the promotion ends. That is a candidate failure path, not a forecast.

The verdict is modify: do not sign the fixed lease on the present case. Fix the shift handover, price a bounded overflow option for peak weeks, and reopen the lease only when ordinary orders repeatedly exceed documented capacity and the cost comparison supports it. The founder assigns those checks to the operations and finance leads before the next lease discussion. Different records could have justified keep or drop. Here the objection mattered because a check changed the immediate action, which the spoken “what if?” alone had not done.

What does keep, modify or drop mean?

Keep means the action still fits the evidence and the stakes after the named checks. Record what survived and what would make you revisit it. A surviving plan is not thereby proved safe or successful; chapter 4 is explicit that surviving criticism adds information without conferring immunity.

Modify changes the claim, the scope, the timing or a safeguard. In the lease example, the founder withholds a fixed commitment and tests a less binding way to handle peaks. Write the replacement action precisely enough that someone else could tell whether it happened.

Drop means the current action has no defensible case under the conditions you have set. Keep the reason, so the same attractive idea does not return later under a new headline. If crucial evidence is missing, say so instead of treating silence as support. A pause or a narrower commitment can be the responsible form of modify; do not write keep simply because the meeting ran out of time.

For a choice that is hard to undo, keep the sheet beside a decision journal written before you act. The journal preserves what you believed when you chose. The sheet shows how you challenged that belief. Later, you can review the reasoning instead of replacing it with a story assembled after the result.

Isn’t this just a pre-mortem?

A pre-mortem assumes the plan has already failed and asks why. On the sheet it is move five of six. Its failure stories generate possible checks; they do not show that the imagined outcome is likely. The full sheet also names the claim, represents the opposing case fairly, looks for contrary evidence, inspects assumptions and records a verdict.

The difference matters when a vivid story takes over the room. Put the story on the sheet as a possibility. Then ask what evidence or early signal would distinguish it from a plan that works. Treat an impressive speech by a devil’s advocate the same way: as a serious lead to inspect, not a vote that ends the decision.

Can a sheet make it safe to disagree?

No form can do that alone. Chapter 10 of The Evidence-Aware Life brings power into the method: a shared process has to ask who risks speaking and who can act on a concern. Assigning a junior colleague the devil’s-advocate role does not remove the cost of disagreeing with a founder. The chapter makes the general point too: a dissent role can invite scrutiny, but it cannot by itself make scrutiny safe.

For a team decision, collect short written views before the senior person’s position is known, and let that person speak last. Allow a private route and an option to pass. Name a decision owner who can act on a serious objection, and tell the person who raised it what happened. Do not promise anonymity when the details would identify the speaker. If dissent has been punished before, find a sponsor with the authority to protect people before you try a public challenge.

The decision owner may still proceed despite an unresolved objection. Record why, what limit applies and when the objection will be revisited. Colleagues do not have to withdraw a concern in order to carry out a bounded decision.

Copy the one-page record

Use one sheet per decision. Keep answers short, but separate evidence already checked from evidence someone only intends to look for. Leave every unknown visible.

Decision and deadline:
Claim the action depends on:
Current supporting evidence:
Cost of being wrong; can the action be reversed?

Strongest fair objection and its support:
Observation that would count against the claim:
Record to inspect / person checking / due date:
Result checked, or "not checked":
Necessary assumption and how to test it:
Possible failure path and early signal:

Verdict: KEEP / MODIFY / DROP
Reason and exact action now:
What remains unknown:
Decision owner / review date / trigger to reopen:

Keep the sheet with the decision. Add later results without erasing the earlier judgment. That lets someone see what changed, rather than assume the final plan was obvious all along.

Try this today (30 minutes)

Choose one plan you are about to commit to. The practice in chapter 4 budgets thirty minutes for the six moves; here is one way to split them. Give the claim, the action and the cost of error five minutes. Spend five more stating the strongest supported opposing case. If you cannot represent it fairly, ask someone who holds it before you declare it answered.

Use ten minutes to name the disconfirming evidence and inspect at least one relevant record you can reach today. Note what you found, what you could not check and who owns the next check. Spend five minutes on the necessary assumption and on one pre-mortem failure path with an early signal. Treat both as questions to investigate, not as evidence that failure is coming.

Use the final five minutes to write keep, modify or drop. State the action that follows, the remaining unknown and the date or event that would reopen the call. If a decisive record cannot arrive before you commit, mark the gap and choose deliberately: delay, reduce the commitment or proceed with an explicit limit. A blank is not supporting evidence. The test is done when the objection has a check, an owner and a recorded verdict, with its open questions in writing. Whether the practice improves the eventual outcome is a separate question.

Cite this:Red-team sheet vs devil's advocate: test the objection.Len P. van der Hof. https://lenvanderhof.com/en/blog/red-team-sheet-vs-devils-advocate/ · Published 27 October 2026.