The decision record for leadership teams
Better decisions,
made together.
Your team gives its honest read in a couple of minutes, before you commit. You decide with the evidence in view, and every big call comes back on the date you set, so the company learns what works.
Time the couple of minutes yourself: answer a colleague's ask in the demo. No signup.
Decision process
67/100
- Status & activity
- People
- Frame
- Alternatives
- Conditions
- Premortem
- Hearing
- Commitment
Input is up to date · 9:20 AM
Run a second assembly shift from October to bring lead time back to 18 weeks
People
5 asked · 4 answered
Conditions for Run a second assembly shift on bay 2 from 5 October
How each condition has been ruled
5 must be true · 2 must not fail
c5Must be true · material
The extra output is sold at list: the book absorbs faster delivery without discounting
Where everyone stands
Elaine Whitfield It holds
David Okafor At risk
Sized against today's book. If the Chicago contract lands, twelve more machines arrive on top of it.
What is your call?
Midfire is the decision record for CEOs of 20 to 100-person companies: a strategy consultant's method on every decision, augmented by AI, with the team's independent read gathered before the decision-maker commits and a dated look-back that says whether the call was right.
Each step below names the research it rests on, and each claim links to where you can check it.
How it works
From the question to the look-back.
One person runs the decision and everyone else answers a few questions. Midfire drafts the analysis for you to edit, keeps the record, and brings the call back when it is due. Here is one decision from the demo, at Hartwell Machines, an 85-person manufacturer.
01 · Frame
Frame the call in minutes.
Write the decision in a sentence. Midfire drafts the frame for you to edit: the question, the paths worth weighing, keeping things as they are included, the line you will not cross and the date you need to decide by.
- Every path gets the same scrutiny, the status quo too1
- Everyday calls take a few minutes; big bets get the full treatment
About half of the decisions Paul Nutt tracked over two decades were not in use two years later. The ones that lasted had weighed more than one option. Nutt, Why Decisions Fail, 2002. Read the guide
1The question
Run a second assembly shift on bay 2 from October and bring the standard lead time back from 22 weeks to 18?
Decide by Friday 21 AugustJudged for the whole company, this year2The paths
- A second assembly shift on bay 2 from 5 October, nine fitters, no new bay
- The June package: a second machining shift and a third assembly bay, five months
- Hold at 22 weeks and ration the order bookStatus quo
3Hard lineNo standard order slips because the shift was announced before it was staffed.Figure 1 The frame, drafted for you to edit.
- 1The question, in one sentence.
- 2Every path, keeping things as they are included.
- 3The line the decision may not cross.
02 · Ask
Hear from your team before they hear from you.
Ask the people whose read matters about the claims that decide it. Each gets one email, answers in a couple of minutes from any device, and learns no new tool.
- Nobody sees another answer until their own is in2 Try it in the demo
- "I don't know" is one click, so the honest answer is the easy one
- Silence stays apart from never having been asked
Once people can see each other's answers, their estimates converge and the group loses what each person knew. Collecting every judgment before anyone hears another's is what Kahneman and his co-authors call decision hygiene. Lorenz and colleagues, PNAS, 2011; Kahneman, Sibony and Sunstein, Noise, 2021. Read the guide
Midfire
Your view
Run a second assembly shift from October to bring lead time back to 18 weeks
Meg Hartwell asked for your view on this decision in progress.
1What is needed
• Weigh in on 2 conditionsAbout a minuteHoping to decide by Friday 21 August
Your name and the time go on the record with whatever you say.
Asked by Meg HartwellCondition 1 of 2
The 18% premium plus cover costs about £90,000 a year and keeps machine margin above 46%
2 You · Where do you stand?
It holdsAt riskI don't knowYou understand the claim but cannot tell whether it is true. The most useful answer when it is the true one; nothing is held against it.The claim is off3Nobody else's answer is on this screen. They open to you once yours is in.
Figure 2 The ask, from both sides.
- 1One email that says what is needed and how long it takes.
- 2One condition at a time, four ways to answer.
- 3Nobody else's answer until yours is in.
03 · Test
See what would sink it while you can still act.
Every claim the path rests on sits in one view, with what each person said and the documents checked against it. Before you commit, the strongest case against your path gets a hearing.
- Every path gets a premortem: the ways it could fail, written down while there is time3
- Disagreement shows up early, with names and reasons
- A checked document leaves a receipt: the passage, the file, the date Read the commitment
- You make the call on each condition: met, risk accepted or not applicable
Imagining that a plan has already failed raises the number of reasons people find by about 30 percent, the finding behind Gary Klein's premortem. Mitchell, Russo and Pennington, 1989; Klein, Harvard Business Review, 2007. Read the guide
c5Must be true · material
The extra output is sold at list: the book absorbs faster delivery without discounting
1Where everyone stands
Elaine Whitfield It holds
The book is 7.1 months, with two 24-week contracts inside it. Faster delivery is sold before it is built.
David Okafor At risk
Sized against today's book. If the Chicago contract lands, twelve more machines arrive on top of it.
2What is your call?
Risk acceptedMetNot applicable3The hearing · the case against this path
Undersized on the day the Chicago contract is signed
The shift is sized to today's book. Twelve more machines a year would put the standard quote back past 22 weeks within a quarter.
RefuteConcede & mitigateAccept riskFigure 3 The test.
- 1Where everyone stands on one condition, with names and reasons.
- 2Your call on the condition.
- 3The strongest case against the path.
04 · Commit and look back
Commit, then find out if you were right.
When you commit, your predictions and tripwires are sealed with the date. The review comes due by email and asks two separate questions: did the call hold up, and did it work out.
- A good call with a bad outcome is recorded as exactly that4 See the record in the demo
- Tripwires say in advance what would make you change course
- Every decision exports as plain files, any time Read the commitment
Judging a call by how it turned out, which Annie Duke calls resulting, teaches the wrong lessons. The forecasters who improved were the ones who wrote down a probability and scored it against what happened. Duke, Thinking in Bets, 2018; Tetlock and Gardner, Superforecasting, 2015. Read the guide
1 Committed · sealed 21 Aug
70%Standard lead time quoted at 18 weeks by 31 March 2027
Scored against what happened, at the review2Tripwires
- Fewer than six fitter offers accepted by 18 September: the start moves to November
- Any late-turn build fails final inspection: the shift stops until engineering signs off
18 Sep first check4 Jan three months31 Mar the horizonGuarantee 18-week delivery on Q1 orders, capped at ten slots
- said 50%Ten slots subscribed at list priceit happened
- said 70%90% of guaranteed orders ship within 18 weeksit didn't
3Did it hold up? mostlyDid it work out? worse than expectedA delivery guarantee is a bet on the slowest constraint in the building. Sequence the fixtures before you sell the promise.
Figure 4 The commitment, and the look-back on an earlier call.
- 1The prediction, sealed with its probability.
- 2Tripwires, set in advance.
- 3The call and the outcome, judged apart.
Built for a leadership team
Easy for everyone you ask.
One person carries the decision. Everyone else answers a few questions from an email, and anyone in the company can read what was decided and why.
The decision maker
Frames the call, asks for input, hears the case against, commits and owns the review.
The seat you pay for
The people you ask
Answer one condition at a time from an email, in a couple of minutes, with no tool to learn and no meeting to attend.
Free, unlimited
Everyone else
Reads what was decided, why, and what was expected, in the folders open to them.
Free, unlimited
Who's involved
- Elaine WhitfieldFinance Director2 conditions (c4, c5)Answered both
- Stuart BellOperations Director3 conditions (c1, c2, c3)Answered 1 of 3
- Tomasz ZielińskiHead of EngineeringPremortemTwo causes given
- Gareth LlewellynField service1 condition (c7)Not yet
Meeting viewCondition 2 of 7
Eight experienced fitters can be hired in Redditch at the 18% premium by the end of September
Independent by default
Each person's answer stays sealed from the others until theirs is in, so the first opinion voiced does not become everyone's.
Meeting view
Put a condition on the shared screen with counts and no names, and the discussion starts on the question.
Who is involved
See who was asked what, who has answered and who has not, on every decision.
What was decided, in one place
Publish a decision's brief to the company. Collections keep every committed decision, month by month.
Follow-through that holds
Review dates come back by email, each review checks the tripwires, and a daily or weekly digest keeps everyone current.
Drafts that know your company
Describe what you do, your strategy and your competitors once, and Midfire frames every decision with it in mind.
The dice tax
What better decisions are worth at your size.
Published research puts a price on deciding without a routine: work aimed the wrong way, and calls that quietly fail. Move the dials to see it for a company like yours.
About half of organizational decisions fail.
Nutt, 400+ decisions"90% sure" managers are right about half the time.
Russo & SchoemakerExperts judging identical cases differed by a median 55%.
Kahneman et al.
The gap between the average company and the top quartile, from Bain's and Paul Nutt's published research. It describes companies, never the effect of a tool. Open the full calculator, with every source and assumption.
Trust
Your decisions stay yours.
What happens to your data, in plain words. The long form is on the trust page, in the privacy policy and in the terms.
Documents are read, never kept
A document checked against a condition leaves a receipt on the record: the passage, the file name, a hash and the date. The file is discarded.
Read the commitmentNo training on your content
Your decisions and documents go to the model provider to answer your request, and to nothing else.
Read the commitmentYour record leaves as plain files
Any member can export what they can open, from Settings, at any time.
Read the commitmentEvery change has a name and a time
Answers, calls and edits are attributed and kept in each decision's history.
Read the commitmentYou decide who sees what
Folders set who reads. People you ask see the decision they were asked about.
Read the commitmentHosted in the United States
On United States providers, each listed with what it receives on the subprocessors page.
See the providers
Guides
How good leadership teams decide.
Short guides for CEOs of 20 to 100-person companies, each answering one question with the research named.
- GuideHow do you run a premortem with a leadership team?The premortem in forty minutes, the research that says it works, and the two mistakes that make it a formality.Read the guide
- GuideWhy should people give their read on a decision before they hear the CEO's view?The order in which people speak decides what a meeting learns. The studies behind independent input, and how to run it without adding a meeting.Read the guide
- GuideWhat is a kill criterion, and how do you write one?A kill criterion is a state and a date, agreed before the bet is placed, that tells you to stop. Why it works, how to write one, and the tripwire that goes with it.Read the guide
Questions
What CEOs ask before they try it.
The longer answers on data are on the trust page, and the words a decision record uses are in the glossary.
Does my whole team have to learn a new tool?
No. One person runs the decision. Everyone else gets an email with a few specific questions and answers them in a couple of minutes, one at a time, before seeing anyone else's view. "I don't know" is a complete answer.
We already keep a decision log in Notion. What does this add?
A log records what you decided. Midfire records what would have to be true for the call to work, who thought so before they heard your view, and whether it turned out right when the look-back comes due. Keep the log if you like it; the record is a different thing.
Does the AI make the decision?
No. It drafts the conditions the path rests on and the strongest case against you, and it grades each condition against the evidence on the record. A named person commits, and the record says who.
What happens to our documents and data?
A document you check against a decision is read once and never stored; the record keeps a receipt. We do not train models on your content. Your record exports as plain files any time. Hosting and subprocessors are in the United States, listed on the trust page.
What does it cost?
$59 per decision-maker per month billed annually, or $79 monthly. People who read or give input are free and unlimited. Each decision-maker adds 50 decisions a month, from quick calls to big bets, to a pool the whole company shares.
Is it built for a company our size?
It is built for companies of 20 to 100 people with a leadership team of four to ten. Below about fifteen people the founder mostly decides alone and the room is small. Above about 150, single sign-on and audit exports come up, and we add those when a customer asks.
Can I see it before I commit to anything?
Yes. The demo runs three fictional companies with real decisions, and it needs no signup. To run it on your own company, bring a real decision to a session with us: onboarding is hands-on today, and the first decision is done together.
Bring your next big decision.
Bring a decision you have to make this month. We frame it with you and send your team its first questions. Or step into the demo first: three companies, real decisions, no signup.