Red Teaming a Decision: Pre-Mortems, Devil's Advocates and More

The most dangerous moment in a big decision often comes right after the plan starts to feel finished. Momentum builds, the people who wrote the plan grow attached to it, and doubts get quieter just when they matter most.

Red teaming is the deliberate practice of attacking your own plan before the world does. The UK Ministry of Defence defines it as "the independent application of a range of structured, creative and critical thinking techniques to assist the end user make a better-informed decision or produce a more robust product." You do not need a military staff to use it. Most of the techniques fit into a single meeting.

This guide covers four of them: a pre-mortem that traces failure back to its root, an assumptions check, devil's advocacy and a simple rule for who speaks when. It closes with how to make sure the challenge actually changes the decision.

Why plans need an adversary

In his September 2007 Harvard Business Review article on the project premortem, Gary Klein wrote that one reason projects fail is that "too many people are reluctant to speak up about their reservations during the all-important planning phase." His remedy was to make it safe for dissenters who know the project well and worry about its weaknesses to speak up.

The third edition of the Ministry of Defence's Red Teaming Handbook (June 2021) takes a practical line. Formal, independent red teams are valuable, it says, but they take time and resources to set up. So the handbook promotes a "red team mindset": everyday habits that individuals and teams can use routinely, including quick techniques for time-pressured situations. It lists what these approaches can do, including uncovering hidden biases, challenging assumptions, identifying flaws in logic, widening the information search and identifying different options.

Run a pre-mortem that reaches the root cause

In Klein's premortem, team members start by assuming that the project they are planning has just failed. The Red Teaming Handbook sets out a step-by-step pre-mortem that works backward from each failure to a decision you control:

1. Assume the plan has failed spectacularly.
2. Define what failure looks like, such as the project being cancelled.
3. Name the immediate causes of that failure. The handbook suggests between two and six is a good number, but there is no limit.
4. For each cause, build a chain backward: what event would have caused this, and what would have caused that?
5. Stop when you reach an action or decision point in the plan. That is a root cause under the decision-maker's control.
6. Repeat for each immediate cause.
7. Decide what to change in the plan to address each root cause.

For example, imagine a regional distributor planning a second warehouse. Failure: the site closes within two years. Immediate cause: volume too low to cover the lease. Working backward: a major customer did not shift orders to the new site, because nobody asked them to commit, because the plan assumed current customers would follow. The root cause is the decision to sign a long lease before confirming volume. The fix might be letters of intent first, or a shorter lease with a renewal option.

Check the assumptions underneath

Every plan rests on things people believe without testing. The US government's Tradecraft Primer on structured analytic techniques (March 2009) defines a key assumption as "any hypothesis that analysts have accepted to be true and which forms the basis of the assessment." Its stated goal is not to abandon assumptions but "to make them explicit and identify what information or developments would demand rethinking them."

The Red Teaming Handbook's version of an assumptions check is simple:

• Write out a short description of the situation or plan.
• Highlight the explicit assumptions, the ones stated outright.
• Highlight the implicit ones. Phrases such as "typically," "generally," "experience has shown" and "would have to be" are clues.
• For each assumption, ask how valid it is and what the impact would be if it were wrong.

The handbook warns that past experience misleads when people assume the current situation repeats an earlier one or is unlikely to change dramatically. In volatile conditions, that is exactly the assumption to test first.

Assign a devil's advocate, and mean it

In the handbook's description, once a plan is developed, the devil's advocate argues that it will fail and, more importantly, identifies the reasons why. The handbook recommends giving the role to someone who was not deeply involved in building the plan, rotating it among people, and keeping the tone constructive so criticism is not given or taken personally.

The handbook's sharpest warning is about token challenge. If the role gets only lip service, "the flawed process can instil more confidence in the plan because it has apparently been tested, when in reality it has not."

The Tradecraft Primer adds a useful trigger for when to use it: when a group has worked on an issue for a long time, "it is probably wise to assume that a strong mind-set exists" that deserves closer scrutiny.

Fix the room before you fix the plan

None of these techniques work if the most senior voice sets the answer in the first five minutes. The Red Teaming Handbook offers a rule you can use tomorrow: "everyone speaks once before someone speaks twice, seniors speak last." It notes that junior team members are often better informed on the subject, yet deference and pressure to agree can silence them.

You can use the rule once at the start of a discussion, for each new topic, or for the whole meeting. If you use it throughout, vary the order of speakers each round and let people pass.

Make the challenge count

A red team exercise is only as good as what happens afterward. For formal red teams, the handbook says the sponsor must be prepared to accept and act on the findings rather than treat the work as an academic exercise. It also lists real downsides of outside red teams: they can be resented by the planning team, time-consuming to set up and resource hungry.

For a small organization, a light version works well:

• Time-box the session, for example 60 to 90 minutes for a major decision.
• Invite one or two people from outside the planning group, such as a trusted peer, adviser or colleague from another department.
• Record each challenge raised and the response: changed the plan, added a mitigation, or accepted the risk knowingly.
• Revisit that record at the first review point after the decision.

Key takeaways

• Red teaming means deliberately challenging your own plan with structured techniques before you commit.
• A thorough pre-mortem traces each imagined failure back to a decision point you control, then changes the plan there.
• An assumptions check makes hidden beliefs visible and asks what happens if each one is wrong.
• A devil's advocate needs real authority to be heard; token challenge can create false confidence.
• Let everyone speak once before anyone speaks twice, with the most senior people last.

Sources

• UK Ministry of Defence, Red Teaming Handbook, 3rd Edition (June 2021)
• UK Ministry of Defence, Red Teaming Handbook, 3rd Edition, PDF (June 2021)
• US Government, A Tradecraft Primer: Structured Analytic Techniques for Improving Intelligence Analysis (March 2009)
• Gary Klein, Performing a Project Premortem, Harvard Business Review (September 2007)