How to Write a Decision Brief (Free Template)

A decision brief is a small document for a large responsibility: making it possible for someone to understand what was decided, why it was decided, and what would make the decision wrong.

It is not a longer meeting agenda. It is not a strategy deck with a new title. A useful brief compresses the decision without hiding the assumptions underneath it. A board member should be able to read it in a few minutes and ask a better question. Six months later, the decision owner should be able to revisit the reasoning without relying on memory.

This guide explains the anatomy of a one-page decision brief, walks through a worked example, and includes a free template you can copy and adapt.

TL;DR

• Begin with one decision, one owner, and one deadline.
• Separate context and evidence from assumptions and interpretation.
• Include real alternatives, not only the option you already prefer.
• Preserve agent and stakeholder positions before compressing them into a recommendation.
• Name risks, early signals, confidence, and what would change your mind.
• Copy the template below or run a focused Decision Sprint when the decision is too important to prepare alone.
• Thornfield Partners had 12 partners adopt the format with an average review time under one hour. Read the case study.

The anatomy of a one-page brief

Decision: Write the choice as an action, not a topic. “Should we grow?” is a topic. “Should we hire two enterprise account executives in Q4?” is a decision. Add the owner and deadline so responsibility cannot dissolve into the group.

Context: Explain what changed and why the decision matters now. Include only context that can change the choice. A good context section is not a company history.

Options: List at least three credible paths, including delay or no action when those are real choices. If the document contains only the preferred option, it is advocacy, not deliberation.

Agent and stakeholder positions: Preserve the distinct arguments. Strategy may favour speed, finance may object to runway, customer voice may flag trust, and operations may question capacity. This section prevents synthesis from erasing useful disagreement.

Risks: State the failure mode, the earliest signal, and the mitigation. “The market could change” is not a useful risk. “Three target accounts do not respond to the revised positioning within 30 days” is observable.

Recommendation and confidence: Make the call, explain the decisive reasoning, and say whether confidence is low, medium, or high. Confidence is not a probability claim; it is a signal about how much uncertainty remains.

Next steps: Assign owners and dates. Include a review date so the decision becomes a learning loop rather than a document that disappears after approval.

A worked example

Imagine a Series A company deciding whether to add a second sales region while it has nine months of runway.

The weak brief says: “We should expand to the US because the market is larger and competitors are moving. We will hire a salesperson and evaluate results later.”

The stronger brief says the decision is whether to fund a 90-day demand test, not whether to commit to a full market entry. It lists three options: launch immediately, run a capped test, or stay focused on the existing region. Finance notes that a full launch could consume three months of runway. Operations notes that the founder cannot own another GTM motion while hiring a product leader. Customer voice says interest from three inbound accounts is not yet evidence of repeatable demand. The contrarian asks whether competitors' announcements are creating urgency without proving a buyer need.

The recommendation becomes: run a €30k, 90-day test against 200 defined accounts, with three paying customers at current ACV as the threshold for a larger investment. Confidence is medium. The key uncertainty is whether the current positioning transfers to US buyers. The review date is fixed before the test begins.

That brief is useful because the choice is explicit, the alternative is real, the objection is preserved, and the next decision has a trigger.

Common mistakes

Writing a recommendation before the question: If the document begins with “we should,” the rest often becomes evidence collection for a conclusion. Write the decision and options first.

Confusing information with context: A list of market facts is not a decision brief. Include the evidence that changes the choice and identify what remains unknown.

Using fake alternatives: “Do it now,” “do it later,” and “do it better” are not always three options. A credible alternative should have a different resource commitment, risk profile, or strategic logic.

Flattening disagreement: A sentence such as “the team agreed that…” can hide the most important information. Keep the strongest objection and say whether it changed the call.

Removing uncertainty: High confidence is not the same as zero risk. Write what would change your mind and when you will look again.

Ending at approval: A decision is not complete when the document is signed. The owner, next step, signal, and review date are what turn it into an operating system.

How an AI council can help

An AI council is useful when the decision brief needs perspectives that are hard to assemble at short notice. Give the council the context, constraints, options, and your current preference. Ask specialised agents to write their positions before a synthesiser creates the recommendation.

The output should not replace the brief's owner. It should make the hidden work visible: the financial objection, the customer risk, the operational dependency, and the test that would resolve the biggest uncertainty. Compare that workflow with ChatGPT vs NeuroAgents and the best AI tools for founders making strategic decisions.

Thornfield Partners is a useful proof point because adoption happened at the partner-group level. Twelve partners used the same brief format and average review time fell below one hour. The benefit was not automation for its own sake; it was a shared starting point for human judgment.

Free decision brief template

Use the inline template below for your next decision. Keep it short enough to read, but do not remove a field simply because the answer is uncomfortable. If a section is unknown, write “unknown” and assign someone to investigate it.

Bring the completed brief to the decision meeting before opening a slide deck. Give attendees a few minutes to read it silently, then start with the strongest objection rather than the recommendation. Record any material change to the options or assumptions directly in the brief, assign the next action before the meeting ends, and calendar the review date. The template works when it becomes the shared record of a decision—not when it becomes another document to polish.

Frequently asked

How long should a decision brief be? Start with one page. A complex decision may need appendices or source links, but the core question, options, recommendation, risks, and next steps should remain scannable.

Who should write the brief? The decision owner should own the question and recommendation. Others can contribute evidence, positions, and objections, but ownership should not be outsourced to the loudest meeting participant.

Should a decision brief include every piece of research? No. Include the evidence that changes the decision and link to supporting material. The brief is a reasoning artifact, not a document repository.

When should I use a Decision Sprint? Use one when the decision is imminent, cross-functional, expensive to reverse, or difficult to challenge alone. The Decision Sprint produces a structured Decision Audit Trail you can share and revisit.

DECISION BRIEF

Decision:
Decision owner:
Decision deadline:

Context:
What changed? Why does this decision matter now?

Options considered:
1.
2.
3.

Agent / stakeholder positions:
Strategy:
Finance:
Customer:
Operations:
Risk:
Contrarian:

Key assumptions:
1.
2.
3.

Risks and early signals:
Risk:
Signal:
Mitigation:

Recommendation:
Confidence: low / medium / high
What would change our mind:

Next steps:
Owner:
Due date:
Review date: