We plan for weeks, and the plan never meets a customer. We read twelve articles on starting, highlight three books on execution, and build a color-coded board. Here is the contrarian claim: most of that work protects us from data, and we call the protection thoroughness.
A 2-minute timer treats overthinking as a workflow bug. Under real pressure, the timer loses, and the old loop returns. We over-analyze because being wrong in public threatens our self-concept.
The fix costs one question per decision and 24 hours per reversible call. The evidence sits in a log I kept for three months. I tracked 94 decisions and recorded two numbers for each: the cost I feared and the cost I paid.
The average real cost of a failed reversible decision: under three hours, zero lasting damage. The ratio of real cost to feared cost: roughly 1 to 300, a gap I call the 300x Rule. The number changed how I decide, and the framework came after.
What is the research spiral actually buying?
The spiral runs a permission machine, and the machine sells the feeling of progress. The data that would move a decision exists in the first five minutes of trying. Everything after minute five buys a better story for the delay while the decision stays put.
Count what a spiral consumes: tabs, highlights, boards, and a doc of pros and cons. Every closed tab feels like a deposit. Nothing lands in the decision itself.
Every new article is another deposit into a machine that manufactures the feeling of progress. The machine never pays out in certainty. It pays out in alibis.
“Being thorough” is the story. “Doing this right” is the story. The smarter we are, the better the invoice looks.
Research becomes the alibi, and the alibi taxes our future self.
The receipt: at 11:47 PM on a Tuesday, I sat between two nearly identical domain names. Forty minutes went into toggling, a pros-and-cons doc, and a text to a friend. He didn’t reply.
I closed the laptop and promised fresh eyes in the morning. Three weeks later I owned neither domain, and the project hadn’t started. The decision was never about the domain.
It was about making the project real enough to fail. The pro-con list gave the fear a home. That is the product the machine sells.
What does a week of waiting cost?
A week of waiting costs the learning that same week would have produced. Ship a rough version in week one, then iterate for five weeks against real signals. Plan for six weeks and ship something polished, and the project banks zero weeks of data.
By week six, the shipper has made five course corrections from live signals. The planner has made zero. The two sit a fundamentally different distance from product-market fit.
The compound interest runs on information. A rough landing page in week one produces a click number by Friday. A perfect one in week six produces a guess.
The timeline slips a week, and the feedback loop slips with it. One person accumulates signals. The other accumulates certainty about a plan.
Planning feels like progress. Shipping is progress.
Which decisions deserve the slow treatment?
Only irreversible decisions deserve the slow treatment, and they turned out to be rare. Before any research, ask one question: is this reversible? A wrong call you can undo inside a week doesn’t need an analysis phase to justify action.
The swap matters. “Am I ready?” has no clear answer. “Is this reversible?” almost always does.
A lookback made the rarity plain. I pulled 20 decisions I had agonized over during six months. Seventeen were fully reversible within a week if they went wrong.
I had treated all 20 like the 3 that weren’t. The three irreversible calls got the slow care, and they needed it. The full filter deserves its own write-up.
The short version: classify first, then act. The filter replaces a subjective judgment with an objective one. Most decisions we agonize over can simply be undone.
Three entries carry the pattern:
Wrong email platform. Feared three weeks to rebuild. Actual cost: four hours.
Too-low pricing on the first offer. Feared credibility damage. Actual cost: one email to existing buyers.
Rough first blog post. Feared reputation harm. Actual cost: zero, and the ones who noticed said “glad you started.”
The fear counted in weeks. The receipts counted in hours. The fear is real, and the cost estimate is fiction.
This is one person’s log, so run the same tracking. Two weeks of feared costs against actual costs settles it. The ratio will surprise you.
What does shipping measure that planning cannot?
Shipping measures four datasets that no planning process can touch. Surveys capture stated preference, and buyers at checkout reveal actual preference. Estimates misread bottlenecks and capacity, the emotional data lives in the body, and only contact with reality can measure it.
Ask people what they want and they will tell you. They also won’t buy it. The gap between stated and revealed preference is where most projects die.
The imagined bottleneck is the landing page. Build it in an afternoon, watch nobody click, and find the real bottleneck in the offer. That data appears only after the page exists.
A project estimated at two weeks takes four, or it takes one. Estimates are guesses. Execution is measurement.
How does it feel to put our name on something imperfect, or to watch someone pay us? Planning cannot simulate the answer. The body produces it only after contact.
The writing habit test ran all four measurements at once. Three weeks of platform research had produced nothing publishable, so I classified instead.
Platform choice: reversible. Content pillars: reversible. Logo at zero readers: irrelevant.
Each decision got the simplest option and a 24-hour cap. The first post went live the next evening, rough, imperfect, live. Six weeks later, 12 posts existed.
Three resonated harder than anything the planning phase predicted. Feedback from those three set the actual content direction. Platform choice took one afternoon instead of three weeks.
Three weeks of research produced zero posts. One afternoon of decisions started a run that produced 12 posts in six weeks.
What happens when a fast call fails?
Some fast calls flop, and the flop invites the oldest voice: we should have planned more. Without a protocol, one bad result tints every later decision, and the spiral returns with evidence. Three written lines convert the outcome into a data point instead of a verdict.
- What the failure taught that planning couldn’t. The data-only-from-shipping column usually surprises.
- What to change next time. One adjustment, the smallest one that addresses the failure.
- The real cost in hours or dollars. An afternoon, a few dollars, one slightly embarrassing post. The fear outran the reality almost every time.
The email platform flop shows line three at work. The rebuild everyone feared at three weeks took four hours. The log entry reads shorter than the worry did.
The protocol converts failure from a reason to stop into a reason to continue.
Chronic planners identify as people who think things through. Experimenters identify as people who run tests. Both respond to uncertainty, and the responses point in opposite directions.
Most advice treats bias for action as a behavior change. In practice it’s an identity change, one that assembles from evidence rather than mood.
James Clear argues in Atomic Habits that identity precedes behavior: identify as a runner, and running follows. My 94-decision log records the opposite order. Behaviors ran first, and the identity updated to match.
I didn’t decide to become an experimenter and then act. I acted 94 times, survived 94 times, and noticed the belief had shifted on its own. We act first, and the identity follows the data.
The 30-day protocol runs on the same discipline. Every decision gets the reversibility check, and reversible calls act within 24 hours. Each result earns one logged sentence.
No additional research. No second opinions. No sleeping on it.
Irreversible calls take all the time they need. The log keeps five columns: date, decision, reversible, action taken, result. At day 30, read it start to finish.
Almost every decision will read as reversible, and the fast outcomes will match the agonized ones. That reading is the evidence. We stop trying to act faster and start holding proof that faster works.
Reality check
The 1 to 300 ratio comes from one log, mine. The protocol costs one sentence per decision for 30 days before the evidence feels real. Irreversible calls still deserve the slow treatment, and no ratio shortcuts those.
Act first, and let the identity follow the data.