A decision is a call the team makes. It can happen at any stage, and it always answers one of four questions: Are there new situations? Does the situation require new concepts? Can the situation be improved? Is the situation worth improving?
For example, after testing, the SecondBank product lead decides to test a points progress bar because new cardholders can't see what their points are worth. That decision connects everything before it, the ask, the situation, the need and the evidence, to the work that comes after.
Why it matters
When decisions aren't recorded, teams relitigate them. Months later nobody remembers why a direction was chosen, who made the call, or what evidence it rested on, so the same debate starts again or the decision quietly unravels.
A recorded decision shows its lineage: why it exists, what shaped it, what limits it, and what supports it. Leaders can see at a glance whether a call rests on tested evidence or on untested assumptions and internal opinion, and step in before the team commits.
Write down what was decided, why and who owns it, link it to the evidence behind it, and check that it's ready before the team commits.
How we write one
Every decision is written as one sentence that names the choice, the reason, and the owner:
We decided to make a choice because a reason**; owner** a person.
We decided to test a points progress bar because new cardholders can't see what their points are worth; owner product lead.
Naming the reason makes the decision easy to revisit when evidence changes, and naming one owner makes it clear who can reopen it.
How it differs from similar ideas
-
An ask is the question. A decision is the answer to it.
-
A recommendation is a suggested action from testing. A decision is what the team actually chose, which may follow the recommendation or not.
-
A bet is a kind of decision: a high-stakes product or marketing call that affects workflows, made in Bets, the last block in Focus. Every bet is a decision, but most decisions aren't bets.
-
A watch item is something to monitor. It can reopen a decision, but it isn't one.
A good decision
-
Answers one of the four decision questions
-
States the choice and the reason
-
Has one named owner
-
Traces back to an ask and a business goal
-
Is supported by tested evidence from customers or participants
-
Says what would reopen it
What happens to it
-
Proposed. A choice is put forward in a meeting, review or recommendation.
-
Made. The owner makes the call, and Ray records the choice, reason and evidence.
-
Committed. The team acts on it: work starts, budget moves or workflows change.
-
Reopened. A watch item, result or new evidence calls it into question, and it goes back for review.
-
Confirmed or reversed. Outcomes show it worked, or a new decision replaces it and the old one is kept for history.
For Ray
This part tells Ray how to find decisions in an input, such as a review meeting transcript, a decision log, a findings readout, a roadmap update or an email from a leader, and how to store them with their full lineage.
How to find one
-
Mark calls that were made. Read the input and mark every statement where someone chose a direction, approved work, or stopped something.
-
Find the choice, reason and owner. Pull out what was chosen, why, and who made or owns the call. If the reason or owner isn't stated, leave it blank.
-
Type it by decision question. Pick the question it answers and its stage: Define (Are there new situations?), Measure (Does the situation require new concepts?), Focus (Can the situation be improved?) or Lead (Is the situation worth improving?).
-
Check whether it's a bet. If it's a product or marketing call made in Bets, in Focus, that affects workflows, mark its kind as Bet and also store a Bet.
-
Trace the lineage. Link the ask it resolves, the business goal and user need it serves, the constraints that limit it, and the signals, findings and metrics that support it.
-
Check readiness. Flag it if it rests on untested assumptions or hunches, has no evidence linked, or is supported only by internal voices.
-
Write the sentence. Use the format: We decided to [choice] because [reason]; owner [person].
-
Fill the record, score it, link it and store it using the rules below.
Cues to look for
-
Decision language: "we decided", "we're going with", "let's move ahead", "approved", "we'll stop"
-
Action items that commit people, budget or time
-
Decision logs, review notes and the last few minutes of a meeting
-
Leaders closing a debate: "OK, let's do it", "go ahead"
What doesn't count
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>If the statement...</p></th><th colspan="1" rowspan="1"><p>Store it as</p></th></tr><tr><td colspan="1" rowspan="1"><p>Suggests an action but nobody has chosen it</p></td><td colspan="1" rowspan="1"><p>Recommendation</p></td></tr><tr><td colspan="1" rowspan="1"><p>Is a question still open</p></td><td colspan="1" rowspan="1"><p>Ask</p></td></tr><tr><td colspan="1" rowspan="1"><p>Is a high-stakes product or marketing call that affects workflows</p></td><td colspan="1" rowspan="1"><p>Decision with kind Bet, plus a linked Bet</p></td></tr><tr><td colspan="1" rowspan="1"><p>Sets a fixed limit, such as a release freeze</p></td><td colspan="1" rowspan="1"><p>Constraint</p></td></tr><tr><td colspan="1" rowspan="1"><p>Is something to keep an eye on</p></td><td colspan="1" rowspan="1"><p>Watch item</p></td></tr></tbody></table>
The record
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 75px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Field</p></th><th colspan="1" rowspan="1"><p>What to fill in</p></th><th colspan="1" rowspan="1"><p>Values</p></th></tr><tr><td colspan="1" rowspan="1"><p>Statement</p></td><td colspan="1" rowspan="1"><p>The one-sentence decision</p></td><td colspan="1" rowspan="1"><p>We decided to [choice] because [reason]; owner [person].</p></td></tr><tr><td colspan="1" rowspan="1"><p>Type</p></td><td colspan="1" rowspan="1"><p>The decision question it answers</p></td><td colspan="1" rowspan="1"><p>New situations, New concepts, Improvement, Worth improving</p></td></tr><tr><td colspan="1" rowspan="1"><p>Stage</p></td><td colspan="1" rowspan="1"><p>Where it was made</p></td><td colspan="1" rowspan="1"><p>Define, Measure, Focus, Lead</p></td></tr><tr><td colspan="1" rowspan="1"><p>Kind</p></td><td colspan="1" rowspan="1"><p>Whether it's a bet</p></td><td colspan="1" rowspan="1"><p>General, Bet</p></td></tr><tr><td colspan="1" rowspan="1"><p>Business goal</p></td><td colspan="1" rowspan="1"><p>The goal it serves</p></td><td colspan="1" rowspan="1"><p>One of the 20 business goals</p></td></tr><tr><td colspan="1" rowspan="1"><p>User need</p></td><td colspan="1" rowspan="1"><p>The need it addresses</p></td><td colspan="1" rowspan="1"><p>Link to User Need</p></td></tr><tr><td colspan="1" rowspan="1"><p>Workflows</p></td><td colspan="1" rowspan="1"><p>Functions it affects</p></td><td colspan="1" rowspan="1"><p>Sales, Marketing, Product, Engineering, Strategy, Operations, Finance, Legal, or blank</p></td></tr><tr><td colspan="1" rowspan="1"><p>Owner</p></td><td colspan="1" rowspan="1"><p>Who owns the call</p></td><td colspan="1" rowspan="1"><p>Name or role</p></td></tr><tr><td colspan="1" rowspan="1"><p>Trigger</p></td><td colspan="1" rowspan="1"><p>What prompted it</p></td><td colspan="1" rowspan="1"><p>Link to Ask, Findings, Watch item or Risk</p></td></tr><tr><td colspan="1" rowspan="1"><p>Rationale</p></td><td colspan="1" rowspan="1"><p>Why it was chosen, and the lineage behind it</p></td><td colspan="1" rowspan="1"><p>Short text plus the lineage statement</p></td></tr><tr><td colspan="1" rowspan="1"><p>Commitment</p></td><td colspan="1" rowspan="1"><p>What the team committed</p></td><td colspan="1" rowspan="1"><p>People, budget, time or work started, or blank</p></td></tr><tr><td colspan="1" rowspan="1"><p>Status</p></td><td colspan="1" rowspan="1"><p>Where it stands</p></td><td colspan="1" rowspan="1"><p>Proposed, Made, Committed, Reopened, Confirmed, Reversed</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>How clearly it was recorded</p></td><td colspan="1" rowspan="1"><p>High, medium, low</p></td></tr></tbody></table>
Ray writes a lineage statement for every decision: This decision exists because [ask and business goal], was shaped by [situation, user need and findings], is limited by [constraints], and is supported by [signals and metrics]. Any blank part of the lineage is a readiness gap.
Scoring confidence
Confidence measures how clearly the decision was recorded: the choice, the reason and the owner. How well it's supported by evidence is judged separately, under readiness.
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Level</p></th><th colspan="1" rowspan="1"><p>When to use it</p></th></tr><tr><td colspan="1" rowspan="1"><p>High</p></td><td colspan="1" rowspan="1"><p>Stated directly by the owner, with a clear choice and reason</p></td></tr><tr><td colspan="1" rowspan="1"><p>Medium</p></td><td colspan="1" rowspan="1"><p>The choice is clear, but the reason or owner is missing or inferred</p></td></tr><tr><td colspan="1" rowspan="1"><p>Low</p></td><td colspan="1" rowspan="1"><p>Implied by action items or later work, with no clear statement that a call was made</p></td></tr></tbody></table>
If any part of the record can't be filled from the source, leave it blank and lower confidence by one level.
Judging readiness
Readiness measures how well the decision is supported before the team commits to it.
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Level</p></th><th colspan="1" rowspan="1"><p>When to use it</p></th></tr><tr><td colspan="1" rowspan="1"><p>Ready</p></td><td colspan="1" rowspan="1"><p>Supported by tested signals from customers or participants, with a full lineage</p></td></tr><tr><td colspan="1" rowspan="1"><p>Caution</p></td><td colspan="1" rowspan="1"><p>Some evidence is linked, but part of the lineage is blank or rests on stated opinion</p></td></tr><tr><td colspan="1" rowspan="1"><p>Not ready</p></td><td colspan="1" rowspan="1"><p>Rests on untested assumptions or hunches, has no evidence linked, or is supported only by internal voices</p></td></tr></tbody></table>
A decision that isn't ready can still be made. Ray records it, flags it, and shows leaders which evidence is missing.
Linking
-
Check for duplicates first. If an existing decision records the same choice by the same owner, add the new source to it instead of creating another.
-
Link what grounds it: the ask it resolves, the business goal and the user need.
-
Link what shapes and limits it: the situation, findings and concepts behind it, and any constraints.
-
Link what evidences it: signals, UX metrics, observations and benchmarks, marked as supporting or contradicting.
-
Link what it triggers: the bet, initiative or workflows it starts, and any watch item that can reopen it.
-
Note reversals. When a new decision replaces an old one, link the two and mark the old one Reversed.
When to store and when to ask
-
Store automatically when confidence is high and readiness is Ready.
-
Add to the review digest when confidence is medium or low, when readiness is Caution or Not ready, when it's a bet, or when it contradicts an open test or an earlier decision.
When to flag
-
A decision rests on untested assumptions or hunches
-
A decision has no evidence linked
-
A decision is supported only by Project Team or Stakeholder voices
-
A decision has no owner
-
A decision has no ask or business goal behind it
-
A watch item linked to a decision has crossed its threshold
Worked example
Input: a findings review meeting transcript, Measure stage.
Product lead: "The concept test was clear. Seven of ten people scrolled right past the balance, and nobody could say what their points buy. We need to fix that, so let's take the points progress bar forward and test it properly. It has to use the existing card component because of the Q1 freeze. I'll own it."
What Ray finds:
-
"Take the points progress bar forward and test it" is the choice; "nobody could say what their points buy" is the reason.
-
"I'll own it" makes the product lead the owner.
-
Moving a concept forward answers Does the situation require new concepts?, so the stage is Measure.
-
"Existing card component because of the Q1 freeze" is a Constraint, linked as a limit.
-
"Seven of ten scrolled past the balance" is an Observation from Participants, so readiness is supported by tested evidence.
Stored record:
<table xmlns="http://www.w3.org/1999/xhtml" style="min-width: 50px;"><colgroup><col style="min-width: 25px;"><col style="min-width: 25px;"></colgroup><tbody><tr><th colspan="1" rowspan="1"><p>Field</p></th><th colspan="1" rowspan="1"><p>Value</p></th></tr><tr><td colspan="1" rowspan="1"><p>Statement</p></td><td colspan="1" rowspan="1"><p>We decided to test a points progress bar because new cardholders can't see what their points are worth; owner product lead.</p></td></tr><tr><td colspan="1" rowspan="1"><p>Type</p></td><td colspan="1" rowspan="1"><p>New concepts</p></td></tr><tr><td colspan="1" rowspan="1"><p>Stage</p></td><td colspan="1" rowspan="1"><p>Measure</p></td></tr><tr><td colspan="1" rowspan="1"><p>Kind</p></td><td colspan="1" rowspan="1"><p>General</p></td></tr><tr><td colspan="1" rowspan="1"><p>Business goal</p></td><td colspan="1" rowspan="1"><p>Drive adoption (User Experience)</p></td></tr><tr><td colspan="1" rowspan="1"><p>User need</p></td><td colspan="1" rowspan="1"><p>Valuable (Impact): new cardholders need to see what their points are worth</p></td></tr><tr><td colspan="1" rowspan="1"><p>Workflows</p></td><td colspan="1" rowspan="1"><p>Product</p></td></tr><tr><td colspan="1" rowspan="1"><p>Owner</p></td><td colspan="1" rowspan="1"><p>Product lead</p></td></tr><tr><td colspan="1" rowspan="1"><p>Trigger</p></td><td colspan="1" rowspan="1"><p>Findings: concept test on point value</p></td></tr><tr><td colspan="1" rowspan="1"><p>Rationale</p></td><td colspan="1" rowspan="1"><p>This decision exists because of the rewards dashboard ask and Drive adoption, was shaped by the rewards tab situation, the Valuable need and the concept test findings, is limited by the Q1 release freeze, and is supported by the 7 of 10 observation and 42% task success.</p></td></tr><tr><td colspan="1" rowspan="1"><p>Commitment</p></td><td colspan="1" rowspan="1"><p>Progress bar concept moves to further testing</p></td></tr><tr><td colspan="1" rowspan="1"><p>Status</p></td><td colspan="1" rowspan="1"><p>Made</p></td></tr><tr><td colspan="1" rowspan="1"><p>Confidence</p></td><td colspan="1" rowspan="1"><p>High: stated directly by the owner with a choice and reason</p></td></tr></tbody></table>
Next step: stored automatically as Ready; a watch item on rewards redemptions is linked so it can reopen the decision later.
Edge cases
-
Silent decisions. Work starts without anyone saying "we decided". Store the decision at low confidence from the action items and ask the owner to confirm it.
-
Decision or bet. If the call commits Product or Marketing in Focus and changes how a workflow runs, mark it as a Bet and add it to the review digest.
-
No owner. "The team agreed" has no owner. Leave the owner blank, lower confidence, and flag it.
-
Reversals. Don't edit the old decision. Store the new one, link it as a reversal, and keep the old one with its original evidence.