Class IRoutine
Two-way door, low cost. Can be undone easily. Ordering supplies, trying a new email subject line, rearranging a schedule.
Time: minutes
DOSSIER
DOSSIER · The system
The Dossier Method: a deep dive into making decisions
Good owners aren't people who never make bad calls. They're people who make decisions the same careful way every time, keep a record, and learn from it. Anyone can learn to decide this way. This page shows you how.
Why you need a system, not just good instincts
Instinct works when you've seen a situation hundreds of times and got quick, clear feedback. Most owner decisions aren't like that. You hire a few times a year, set prices once or twice, and choose a direction even less often. You don't get enough repetitions for instinct to become reliable.
A system fills that gap in three ways. It makes you consistent, so mood and pressure matter less. It makes your thinking visible, so you and others can check it. And it leaves a record, so a year later you can see whether your reasoning was good or you just got lucky.
That record is the dossier: a one-page file for each important decision. Over time your dossiers become the most valuable document in your business, a history of how you think.
■ Triage first
The three case classes
Before you open a file, decide how big the case is. This keeps you from spending a week on a two-minute choice, or two minutes on a choice that could sink the company. You'll see these class labels throughout the site.
Class IISignificant
Costly but reversible. Mistakes hurt but can be fixed. A price change, a new tool, a part-time hire, a marketing push.
Time: an hour or two
DOSSIER
Class IIICritical
One-way door. Hard or impossible to undo. Firing someone, signing a long lease, taking on debt, dropping a major client, a partnership.
Time: days, plus a second opinion
DOSSIER
■ The seven steps
Working the case, step by step
D
Define the question
Most bad decisions start as the wrong question. "Should I fire Alex?" is narrower than "How do I get reliable work done in this role?" Chip and Dan Heath, in their book Decisive, warn about "whether or not" questions because they hide the other options. Rewrite your question until it leaves room for at least two real choices.
- Ask
- What problem am I really solving? What are my options, including doing nothing? When must this be decided? What does success look like in six months?
- Goes in the dossier
- The question, the options, the deadline, the case class, and a one-line success measure.
- Trap
- Solving a symptom. Lost clients may be a pricing problem, a service problem or a sales problem.
O
Take the outside view
Two moves. First, describe the situation in the third person, using your name, so your ego steps back (covered in Third Person). Second, check the base rate: how do decisions like this usually turn out for people in general? Psychologist Daniel Kahneman called this the outside view. Your plan always feels special from the inside. The base rate tells you how often plans like it actually work.
- Ask
- What would a new owner do? What would I tell a friend? How do cases like this usually go for other businesses?
- Goes in the dossier
- One third-person sentence and one base-rate note.
- Trap
- "My situation is different." Sometimes it is. Make yourself say why.
S
Sort the evidence
Put everything you know into facts, assumptions and guesses (see What-If First). Then find your shakiest assumption, the one that would change the decision if it were wrong, and look for the cheapest way to test it before you commit. One phone call or a small trial often beats a week of thinking.
- Ask
- What can I prove? What am I assuming? What evidence would change my mind? How can I test my weakest assumption cheaply?
- Goes in the dossier
- Top facts, top assumptions, and the one assumption you tested or still need to test.
- Trap
- Counting opinions as facts. "Clients love us" is an assumption until you see repeat orders.
S
Run the scenarios
For your leading option, write the best case, worst case and most likely case. Then ask "and then what?" once more. Then run a premortem: assume it failed a year from now and write down why. The key test is survivability. If the worst case would end the business, you need a backup plan or a smaller first step, however unlikely the worst case seems.
- Ask
- What's the best, worst and most likely result? And then what? If this failed, what's the most likely reason? Can I survive the worst case?
- Goes in the dossier
- Three short scenarios, the second-order effect, and the top premortem reason.
- Trap
- Writing a "worst case" that's really just a mildly bad case because the real one is scary.
I
Interrogate your plan
Now question your favorite option like a suspect (see No Yes Man). Write the strongest case against it, ask what would have to be true for it to work, and for Class III cases, get a second person to attack it. If you can't argue the other side well, you don't understand the decision yet.
- Ask
- What's the best argument against this? What has to be true for it to work? Who would disagree, and what would they say?
- Goes in the dossier
- The case against, and who reviewed it.
- Trap
- Arguing against a weak version of the other side so you can win easily.
E
Execute the decision
Make the call. You won't have every piece of information, and that's normal (see Tough Calls for the 70% rule). Then set tripwires, an idea the Heath brothers also describe: signals you choose now that will force you to rethink later. "If we haven't signed three new clients by March 1, we stop the ad spend." Tripwires protect you from sticking with a bad call because you've already invested in it. Last, tell the people affected clearly. How you explain a decision decides how well it gets carried out. For how to say it, see Clear at Work.
- Ask
- What's my call? What signal would make me rethink it? Who needs to hear about it, and how?
- Goes in the dossier
- The decision, one or two tripwires, and who you told.
- Trap
- Waiting for certainty that never comes, or deciding but never telling anyone the why.
R
Review on a set date
Every dossier gets a review date. When it comes, compare what happened with what you expected and grade the decision. Poker player and author Annie Duke calls the common mistake "resulting": judging a decision only by its outcome. A good decision can have a bad result, and a bad decision can get lucky. The Decision Quality Grid below separates the two.
- Ask
- What happened? What did I expect? Did any tripwire go off? Was my reasoning sound with what I knew then? What would I do differently?
- Goes in the dossier
- The outcome, a grid grade, and one lesson.
- Trap
- Skipping the review because the result was good. Lucky wins teach the worst habits.
R · Review tool
The Decision Quality Grid
At review time, place the decision in one square. Over many dossiers, this tells you whether you have a thinking problem or just bad luck.
Good outcome
Bad outcome
Sound reasoning
EarnedDo it again. Note what made it work.
Bad luckKeep the process. Check if a tripwire could have caught it sooner.
Weak reasoning
Dumb luckThe most dangerous square. Fix the process before it fails you.
LessonFind the step you skipped. Write it down.
■ The artifact
What a finished dossier looks like
This is the deliverable of the whole method: one page, every step covered, readable in two minutes. This example is a Class II decision for a small bookkeeping firm.
Dossier No. 2026-014 · Class II
Example
How do we handle growing admin work without burning out?
D · Define
Options: hire full-time, hire part-time contractor, buy software, do nothing. Decide by Dec 1. Success: owner under 45 hours a week by March.
O · Outside view
"A new owner would get help now." Many small firms hire too early and struggle with payroll, so a trial is safer.
S · Evidence
Fact: turned down 6 clients in 3 months. Shakiest assumption: that I can delegate well. Not yet tested.
S · Scenarios
Best: capacity doubles. Worst: salary with no new work. Likely: slow 2 months, then gains. Premortem: "I never trained them properly."
I · Interrogate
Against: full-time locks in cost before delegation is proven. Reviewed by a mentor.
E · Execute
Hire a 90-day contractor. Tripwire: if I'm not saving 8 hours a week by week 6, stop and rethink.
R · Review on Feb 15
Outcome: to be filled in · Grid grade: to be filled in · Lesson: to be filled in
Free download · 4-page PDF
Download the PDF
The Dossier template
Fill it in on screen or print it. Includes the case classes, all seven steps and the Decision Quality Grid.
Rules of the method
- Every Class III decision gets a full dossier. No exceptions.
- No dossier is closed without a review date.
- Tripwires are set when you decide, not after things go wrong.
- Grade the reasoning, then the outcome, in that order.
- Keep every dossier, including the embarrassing ones. They teach the most.
Ready? Open your first file in the Dossier Builder.
Go deeperCase file, step-by-step how-to, common mistakes and questions to work through
Case file: The second truck
A fictional example
Marcus runs a residential HVAC company with four technicians and two vans. Spring is coming, his phone is already ringing, and he's turning away about six service calls a week. A dealer offers him a used van for $38,000, financed at $790 a month for five years. His gut says yes. His bookkeeper says wait.
D · Define. Marcus first wrote the question as "Should I buy the van?" Then he rewrote it: "How do we capture the calls we're turning away this spring?" That opened four options: buy the van, lease one for the season, have techs work staggered shifts with the existing vans, or raise prices and accept fewer calls. He set a deadline of March 1 and a success measure: capture at least four more calls a week by April.
O · Outside view. In the third person: "Marcus is busy for about four months a year and slow for three. A new owner would want to know what the van does in December." The base rate he found from talking to two other contractors: vans bought for a busy season often sit idle half the year.
S · Evidence. Facts: six turned-away calls a week for three weeks, an average ticket of $420, a 45% gross margin. Assumption: the demand lasts all season. His shakiest assumption was that he could staff a third crew, because he only had four technicians, and a third van needs two more people or splits the existing crews.
S · Scenarios. Best case: four extra calls a week at $420 is about $1,680 a week in sales and roughly $756 in gross profit. That covers the $790 monthly payment easily in busy months, but at that pace the full $38,000 price takes about 50 busy weeks to earn back, which is more than two full seasons. Worst case: he can't hire, the van sits, and he pays $790 a month for nothing. Most likely: the van is busy April through August and idle the rest of the year. Premortem: "It failed because we couldn't staff it."
I · Interrogate. His senior tech made the case against it: "The bottleneck isn't vans. It's people. We already have a van that sits every Saturday." That changed everything.
E · Execute. Marcus chose staggered shifts plus Saturday hours for the existing vans, paid at overtime, and posted a job for one technician. Tripwire: if he was still turning away four or more calls a week by May 1 after hiring, he'd lease a van for the summer.
R · Review. By May he'd captured most of the extra calls without a third van. Grade: Earned. Lesson: "Find the real bottleneck before buying capacity."
How to work a dossier, step by step
- Set a timer. Give a Class II decision about an hour and a Class III decision a few sessions over several days. Deadlines stop analysis from dragging on.
- Write the question three ways. Start with your first version, then write a broader one ("How do we…") and a narrower one. Pick the version that leaves room for at least three options.
- List the options, including the boring ones. Doing nothing, doing it smaller, delaying, and doing a test version all count.
- Fill in the outside view before you look at your own numbers. It's harder to be objective after you've fallen in love with the plan.
- Mark every number as a fact or an assumption. A simple "F" or "A" in the margin is enough.
- Write the worst case in one honest sentence, then ask: can we survive it? If not, change the plan, not the worst case.
- Hand the plan to someone else to attack before you decide, using the wording below.
- Decide, set one or two tripwires, and put the review date in your calendar right then.
Say it like this · asking someone to attack your plan
"I'm leaning toward this decision, and I want you to try to talk me out of it. What's the strongest reason this goes wrong? You won't hurt my feelings. You'll save me money."
Common mistakes and how to fix them
| Mistake | Fix |
|---|---|
| Filling in the dossier after you've already decided, to justify the choice | Open the file when the question first comes up, not at the end. If you notice you're defending, go back to step O. |
| Treating every decision as Class III and stalling the business | Classify first. Most decisions are Class I or II. Save the full process for the few you can't undo. |
| Writing a gentle "worst case" because the real one is scary | Ask, "What would I tell a friend could go wrong here?" Write that version. |
| Skipping the review because things turned out fine | Lucky wins teach bad habits. Review every Class II and III decision, good outcome or not. |
| Keeping dossiers in your head | Write them down. A dossier you can't reread isn't a record, it's a memory, and memories edit themselves. |
Questions to work through
- Think of your best and worst decisions of the last year. Which step of the Dossier Method did you skip in the worst one?
- Which of your current open decisions are actually Class I and could be decided today?
- Who in your life will honestly attack your plans? If no one, who could you ask?
- What's one decision you made where the outcome was good but the reasoning was weak? What would have happened with worse luck?
- What tripwire would have saved you from your most expensive mistake?