A roadmap from idea to evidence
Earned Success
Ninety days to make useful work, expose it to judgment, and begin building a record worth trusting.
Make → Test → Correct → Deliver → Repeat
A moving map / 90 seconds
Step inside the little world.
A maker leaves the workshop, meets a failed test, returns to correct the work, and carries the practice toward dependable public use. The return path is part of the journey.
Read the film and follow the route
00:00 — The Little World of Earned Success. A practical journey from idea to evidence. The whole miniature connects a workshop, a test lab, a correction bench, a delivery bridge, and a public pavilion.
00:08 — Choose the change. Give the idea a job in someone’s life. Name the person. Name the useful change.
00:18 — Make it testable. Build a small expression. Write the test before the result.
00:29 — Meet the world. Watch what happens. The test arch fails; keep the failures in the record.
00:41 — Correct the work. Going back is part of moving forward. The maker returns along the coral route, and a revised arch takes shape. Study the mismatch. Make the next attempt better.
00:54 — Deliver again. Make the useful result dependable. Protect the capacity to continue.
01:06 — Earn the next yes. Bring a record others can inspect. A panel moves out for repair while the pavilion remains in use. Dependability includes maintenance.
01:19 — Make. Test. Correct. Repeat. Make the next honest commitment. Explore the roadmap and workbook below.
Put the roadmap to work
Your 90-day workbook is ready.
Take every worksheet into your own project: all six stages, an observation log, a claim check, the evidence gate, diagnostic exercises, 13 weekly reviews, and a day-90 decision page.
Download the 90-day workbookPDF · 34 pages · 187 fillable fields · 4.9 MB
Fill it digitally or print it. Save a local copy before entering answers. Open the PDF
The starting point
Success needs a definition that can survive contact.
Make something useful. Let its consequences correct you. Become dependable enough to do it again.
In The Past Tense Makes Us Answerable, I argued that completed work gives other people something they can question. Now comes the practical question: how do you organize a life, a project, or a business around that discipline?
Start by refusing a definition of success that can be satisfied by appearances. A beautiful artifact can be useless. A busy practice can be untrustworthy. A celebrated project can consume the person responsible for maintaining it.
For this roadmap, earned success means a useful outcome that someone else can verify, produced through conduct you can defend, with enough remaining capacity to continue. Usefulness, trust, and endurance belong together.
The ninety days below organize the work. They do not dictate when another person must buy, understand, forgive, or agree. You control the seriousness of the practice. The response remains something to encounter.
Useful to someone
Name the person, the difficulty, and the change that would matter in their life.
Open to inspection
Keep enough evidence that your account can be checked without borrowing your confidence.
Possible to sustain
Protect the time, relationships, resources, and judgment the next delivery will require.
The 90-day route
Six stages. Six things to bring back.
Move at the speed of evidence. Revisit a stage when its central assumption fails; a calendar date cannot certify readiness.
Choose the change
Bring back a one-page commitment.
Make it testable
Bring back a small working expression.
Meet the world
Bring back an observation record.
Correct the work
Bring back a visible revision.
Deliver again
Bring back a repeatable method.
Earn the next yes
Bring back a decision others can judge.
The schedules, thresholds, and checklists in this guide are my proposed working method. Adapt them to the scale and consequences of your project. All workshop examples and numerical results below are hypothetical illustrations.
Choose the change
Give the idea a job in someone’s life.
“Ideas have people” changes our posture toward authorship. It does not tell us that every impulse deserves a project. The first responsibility is discernment: what relationship are you participating in, and what worthwhile expression can you attempt now?
Name the condition that orients you—fairness, beauty, clarity—then distinguish it from the project. A registration service is an expression you can build. It is not fairness itself. That distinction lets you change the service without having to defend every design decision as sacred.
Choose one person or group and one observable difficulty. Replace “improve education” with a bounded task such as helping a new participant reserve a workshop seat without needing a staff member to interpret the instructions.
Before proposing the answer, watch the existing process and ask people to describe a recent attempt. Where did they stop? What did they do instead? What help did they need? Listen for events rather than compliments about your concept.
Finish the week with a commitment small enough to keep. Include what you will attempt, what you will not claim, the effort you can afford, and the conditions under which you will stop or change course.
Open the commitment worksheet
- Person: Who experiences the difficulty? Who else is affected when you intervene?
- Starting state: What happens now, and what record supports that account?
- Desired change: What would be observably different for that person?
- Boundary: What must remain protected: access, accuracy, consent, time, or an existing commitment?
- Capacity: How much work can you sustain without borrowing from obligations you already made?
- Decision date: When will you inspect the evidence and decide what follows?
Illustrative commitment: “For one workshop, we will test whether new participants can reserve a seat and receive an accurate confirmation without staff assistance. We will preserve a staffed alternative and record every request for help.”
Make it testable
Build the smallest expression that can teach you.
Small does not mean careless. It means narrow enough that you can understand what a result says. A working registration page for one workshop may teach more than a polished platform for every event you might someday offer.
Identify the assumption most capable of defeating the project. Perhaps people do not understand the offer. Perhaps the process excludes someone. Perhaps the thing can be made but cannot be maintained. Choose the next test around that uncertainty.
Write the test before the result. State who will try the work, what they will attempt, what you will observe, and what finding would make you revise it. Record the version. If you invent the standard after watching the outcome, almost anything can be narrated as progress.
AI can help draft, model, code, compare, and organize the attempt. Keep a named human responsible for inspecting the output, verifying important claims, and deciding when it is ready for use. Cheap production is an advantage only when judgment remains attached.
Your first completion is a testable artifact, not a proven outcome. Say which one you made.
Open the first-test worksheet
- Write: “We think this change will help this person complete this task because…”
- Pick a task someone can actually attempt, rather than asking whether they like your idea.
- Define what counts as completion, assistance, abandonment, and an error.
- Keep the first attempt bounded enough to watch closely and repair quickly.
- Name the finding that would make you pause. Do not treat every objection as a request for another feature.
Illustrative test: “New participants will attempt a reservation using the first page. We will record whether they finish unaided and whether the confirmation matches the seat actually reserved. A mismatch in confirmations pauses further use until repaired.”
The central distinction
Let the work be smaller than the ideal.
A particular drawing does not exhaust circle. A procedure does not exhaust fairness. A finished expression gives your present understanding a visible shape.
That is good news for the maker. You can be serious about the relationship without pretending the first expression was sufficient.
Meet the world
Watch use. Record what happened.
Bring the work to the people whose difficulty you named. A friendly audience can improve your morale while telling you very little about whether the work helps the intended person.
Give participants a clear task and explain what you are observing. Let them attempt it without an uninterrupted tour from the maker. Record assistance rather than quietly rescuing the task and counting it as success.
Keep the denominator. “Six completed” means something different when six attempted than when sixty attempted. Keep the failures, the conditions, the version, and the observations that do not fit your preferred explanation.
Compare with the starting process when you can, while naming what makes the comparison imperfect. Different participants, easier tasks, extra attention, and changed circumstances can influence the result. A small observation can guide a revision without establishing a general cause.
Then ask a second question: what did achieving this result cost the people around it? A faster page that produces inaccurate reservations, excludes a participant, or creates invisible staff cleanup has displaced difficulty rather than resolved it.
Open the observation worksheet
- Record the date, version, task, and relevant conditions.
- Count attempts, completions, assisted completions, abandonment, and important errors using the definitions you wrote.
- Separate what you saw from what you infer. “Stopped at the date field” is an observation; “does not care about attending” is an interpretation.
- Collect a participant’s own account of the difficult moment without correcting their experience.
- Write the narrowest defensible finding and the uncertainty that remains.
Useful record: “In this round, six of eight participants completed without help. Two needed assistance with selecting a date. The confirmation matched the reservation in all eight attempts.”
A worked example
Let the evidence set the size of the sentence.
Imagine two small observed rounds of the workshop reservation task. The figures below are invented to demonstrate careful reporting. They are not results from a real service or evidence of an expected improvement.
Even encouraging observations leave questions open. Who attempted the task? Were the rounds comparable? Will the result hold when you are not in the room?
| What was observed | Earlier process | Revised page |
|---|---|---|
| Attempts observed | 8 | 8 |
| Completed without help | 3 | 6 |
| Needed assistance | 5 | 2 |
In this constructed example, every attempt ended in a reservation; some required help. The two rounds are small and are not a controlled comparison. The counts describe these observations, not the effect the revision would cause across a wider population.
A claim the record can support
“In the revised-page sessions, six of eight participants reserved without help. Two still needed assistance. We are investigating the date-selection step.”
A claim that outruns the record
“We solved access to the workshops.” The test did not examine every participant, every access need, or ordinary use over time.
Correct the work
Make the mismatch useful.
Now sort the evidence into three piles: what worked, what failed, and what you still cannot tell. The third pile matters. Uncertainty should not be promoted into either victory or defeat because a meeting needs an answer.
Choose the most consequential mismatch. A reservation confirmation that cannot be trusted matters before a decorative improvement. A participant who cannot use the date selector deserves more attention than another animation.
Record the earlier assumption, the observation that challenged it, the revision, and the next test. Keep the old version or a dated account of it. A correction is a new act; it does not require the earlier one to vanish.
If your previous claim reached people who might still rely on it, put the correction where those people can find it. Private learning cannot repair a public overstatement by itself.
Then run a relevant test again. Changing the work is a completion. Learning that the change helped is a separate completion. Let your language preserve the distance between them.
Open the correction worksheet
- We previously believed: State the assumption plainly.
- We observed: Identify the event or evidence that challenged it.
- We changed: Name the actual revision and its date.
- We notified: Identify the affected audience, when notice is needed.
- We have not established: Preserve the open question.
- We will check: Specify the next task and observation.
Illustrative correction: “We treated the date selector as self-explanatory. Two participants needed help. We replaced it with a clearer control and revised the instructions. That revision is complete; we have not yet tested whether it resolves the difficulty.”
Deliver again
Turn the good attempt into a dependable practice.
One successful attempt establishes that an attempt succeeded. Dependability asks a different question: can the result recur under ordinary conditions without hidden heroics?
Repeat the work across more than one relevant occasion. Include the routine variation your promise actually covers. If every good result requires you to stay up late, personally rescue the process, and conceal the cleanup, the method is unfinished.
Write the smallest operating guide another capable person would need: starting conditions, steps, checks, exceptions, and the person responsible when something goes wrong. A checklist is useful when it makes judgment easier to exercise, not when it pretends judgment is unnecessary.
Keep an account of the whole delivery: preparation, execution, assistance, correction, and maintenance. Decide what you can continue offering at that level of effort. Reduce the promise before you build a reputation for meeting it through exhaustion.
Success here may mean a smaller scope, a slower rhythm, or retiring a feature. Fidelity is not measured by the number of obligations you can accumulate.
Open the repeat-delivery worksheet
- Define the conditions under which the method is intended to work.
- Repeat the delivery and record the assistance or repair each occasion required.
- Write a short handoff, then observe where another capable person still needs your private knowledge.
- Name the exception that stops normal operation and the person who handles it.
- Compare the real effort with the capacity you committed in week one.
- Adjust the scope or method before increasing the number of people depending on it.
Illustrative operating promise: “Reservations are confirmed against the live seat count. If the count cannot be verified, the request is held for staff review rather than presented as a confirmed seat.”
Earn the next yes
Make your record useful to someone else’s decision.
Bring the evidence together in a form another person can inspect: the original commitment, the finished expression, observations, corrections, remaining limits, and the current operating promise.
A portfolio made only of beautiful artifacts asks viewers to infer the rest. A useful record makes the relationship between work and outcome legible. Show what you did, what happened when it was used, and what changed after you learned more.
Then make one proportionate invitation. Ask for another assignment, a limited next engagement, a reader’s serious critique, or a decision about wider use. Make the proposed obligation clear enough that another person can consent to it with judgment intact.
Choose among four actions: continue the present method, revise a specific part, expand within demonstrated capacity, or stop. Expansion is one possible decision. It is not the automatic reward for reaching day ninety.
A failed project can leave a trustworthy record if you tell the truth and keep your obligations. A successful project can leave an untrustworthy record if you overstate the result. Protect what the next person must be able to rely on.
Open the day-ninety review
- What completed change can an affected person verify?
- Which part of the result remains dependent on favorable conditions or your personal intervention?
- What did you get wrong, and what evidence shows the response?
- What effort does the current promise require?
- Who bears the downside if you expand too early?
- What is the smallest next commitment the evidence actually justifies?
Useful final report: “We built and tested a reservation process for this workshop, corrected the date-selection problem, and documented staff review for uncertain seat counts. Here is the record, including the sessions that required help. Our next proposal is limited to another workshop under these conditions.”
The weekly rhythm
A roadmap works through ordinary weeks.
Monday
Choose one finish.
Name the completed act you can reasonably bring back by Friday. Locate the uncertainty that matters most.
Tuesday–Wednesday
Make and expose.
Build the bounded attempt. Put it in the relevant setting. Preserve what you observe.
Thursday
Study and correct.
Compare the result with the intended change. Repair the most consequential mismatch.
Friday
Speak from the record.
Report what finished, what happened, what remains uncertain, and what commitment follows.
“We completed ___. We observed ___. We corrected ___. We have not established ___. Our next bounded commitment is ___.”
The evidence gate
Before you make a bigger promise.
Use these checks before widening the audience, adding another obligation, or asking others to rely on the result.
Check a box only when you can point to the relevant record. The point is not to collect five reassuring marks. It is to discover the missing work before someone else bears its cost.
Completing the checklist prepares a review. It does not establish that a project is successful or that expansion is wise.
Keep the next promise small. Use any unchecked item to choose the next piece of work.
Bring the record to a review. The five checks are present. Ask an affected person to inspect them before deciding whether to continue, revise, expand, or stop.
Use this as a review aid. Keep the evidence and your decisions in your own project record.
When the route gets difficult
Diagnose the missing relationship.
I keep planning and never finish.
Your next commitment may be too large, too protected from judgment, or attached to an identity you are afraid to test. Reduce it to one meaningful act. Set a date. Decide who will encounter it. Bring back a modest completion rather than a larger announcement.
I finish things, but nobody benefits.
Return to the person and difficulty. Watch the existing process. Ask what your artifact changed after it left your hands. More production may be amplifying an incorrect premise. Change the task before accelerating the output.
People benefit, but the work consumes me.
Inspect the delivery you actually perform, including rescue and maintenance. Narrow the promise, clarify the exception path, change the method, or decline another obligation. A practice that destroys its capacity to continue needs correction too.
What to keep
Build a record, then remain available.
At the end of ninety days, you may have a viable service, a better chapter, a more useful practice, or a well-supported decision to stop. The more durable gain is a way of working that does not require you to protect an imaginary version of yourself.
You can receive without claiming ownership. You can make without claiming perfection. You can report a result without enlarging it. You can correct without pretending the earlier act disappeared.
Make something useful. Keep the evidence. Answer the questions. Become capable of the next honest commitment.
That is the roadmap to earned success.
