Industry
Part of Why your own first hundred days rarely go in a straight line
The first hundred days framework: a single question each week
First 100 days framework built as a weekly loop rather than a plan: one question per week, a fixed four-item agenda, and a written escalation rule.
A hundred-day plan written on day one is a guess about days sixty through a hundred made by someone who has not met a customer yet. It will be wrong, and the wrongness will be invisible because the plan will simply be abandoned without a record.
Run a loop instead. The loop is the framework: one question per week, the same four items reviewed every time, and a rule that says when a week's answer forces a change rather than a note.
What to take away
- The unit is the week, not the quarter. Each week owns one question and produces one written answer, even when the answer is that nothing can be said yet.
- The agenda never changes. Four items, same order, so that a drift in one of them is visible across weeks.
- The escalation rule is written on day one and is the only part of the plan that is fixed.
The week's question
One question, chosen on the Monday, answerable by the Friday from what will actually exist by then.
Good questions at this stage are narrow and about behavior. Do the people who signed up in week one still return in week three. Does anyone use the second feature without being told about it. Do the support questions cluster. Does anyone renew when the first cycle ends.
A question fails if it cannot be answered by Friday because the evidence needs more time to accrue. Note the lag and schedule it for the week it becomes answerable. Half of the discipline here is refusing to answer a question early with the wrong evidence, which is what produces the confident wrong conclusions of the first month.
The four standing items
Reviewed every week in this order, whatever the week's question is.
Four standing items, weekly order
- Who arrivedcounts by route, warm and cold apart
- Who stayedof prior arrivals, who remains
- What they askedsupport and question log, grouped
- Who left and what they saidevery cancellation, their words
| Item | What you write down | The failure it catches |
|---|---|---|
| Who arrived | Counts by route, warm and cold kept apart | Reading your own network as the market |
| Who stayed | Of the people who arrived before, who is still here | Growth that is entirely replacement |
| What they asked | The support and question log, grouped | Fixing the loud problem instead of the common one |
| Who left, and what they said | Every cancellation, with a reason in their words | Losing the only group that explains the ceiling |
The fourth item depends on easy cancellation. Treat that as a design rule, not a courtesy. If your offer bills on a recurring basis, the rule on negative option offers covers consent to those charges and stopping them without obstacles.
A hard cancellation also destroys the record: people who give up and dispute the charge tell you nothing. People who cancel in one click can be asked why.
The fourth item gets dropped, and it is the most informative. People who stop stay reachable about a week, then they are gone. One honest answer to one question beats a month of usage data.
The pillar on the first hundred days covers what to ask them and how to hear it.
The escalation rule
Written before day one, in a sentence with a number in it, and not revised afterwards.
Escalation rule at week six
By end of week six, are fewer than the stated number of weeks one and two arrivals still active?
Stop adding features, spend two weeks on conversations
Continue the weekly loop as written
It reads like this: if by the end of week six fewer than a stated number of the people who arrived in weeks one and two are still active, we stop adding features and spend two weeks on conversations instead.
Two properties make it work. It names a count rather than a rate, so a small denominator cannot dress itself up. And it names an action rather than a feeling, so it cannot be satisfied by concern. Choosing a number that means something to your own arithmetic, rather than borrowing one, is the reasoning worked through in market validation metrics.
What each phase of the loop is actually for
The loop runs the same way throughout, but the questions available to it change as time passes.
What each phase can answer
- Weeks one to threeArrival and comprehension only; retention is noise
- Weeks four to eightFirst cohort stays or not; escalation rule fires or not
- Weeks nine to fourteenRepeat behavior visible; re-ask earliest questions of a second cohort
Weeks one to three. Nothing about retention can be answered yet. The available questions are about arrival and comprehension: where people come from, what they expected, where they stop in the purchase path. Anyone drawing a retention conclusion here is reading noise.
Weeks four to eight. The first cohort has been around long enough to either stay or not. This is where the escalation rule usually fires or does not. It is also where the first honest cancellations arrive.
Weeks nine to fourteen. Repeat behavior becomes visible, and the earliest questions can be asked again of a second cohort to see whether the answer holds. A finding that only ever held for the first group of users is a finding about the launch, not about the product.
The write-up that makes the loop worth running
Half a page a week, same shape, in one file that grows downward.
Half-page weekly write-up
- The question
- The answer, or why it cannot be answered yet
- The four standing items, with counts and denominators
- Anything that surprised you
- One line on what changed, including nothing
Read the last four weeks before writing a new one. That single habit is what turns a stack of notes into a framework, because the drift you are looking for is only visible across weeks. A change of two in a count means nothing; the same direction four weeks running means something even when each step is small.
Where the loop stops
At day one hundred, or earlier if the escalation rule fires twice.
The output is not a verdict on the venture. It is an answer to a narrower question: whether the thing keeps people, whether you can reach more of them, and whether the money works at the volume you can actually deliver. Those three, written plainly with their evidence, are what the next decision is made from.
Where the money question is open, the startup cost breakdown published by the SBA is a useful reminder that one-time and recurring costs behave differently. Only the recurring ones bear on whether this works at volume.
If the honest answer to all three is "not yet clear," the correct move is another loop with a narrower audience, not a bigger push. The decision procedure for that is the same one set out in founder readiness, applied later than usual.
Common questions
Why a week and not a sprint of two weeks?
Because the standing items move on a weekly rhythm and a fortnight lets a drift go unnoticed for one full cycle. If a week produces nothing, write that; the empty entry is part of the record.
What if there are almost no users?
Then the counts are tiny and the loop is mostly the fourth item: talking to the handful you have. That is not a degraded version of the framework, it is what the framework looks like at that size. The techniques belong to customer interviews.
Can the escalation rule be changed?
Not during the hundred days. If it turns out to be wrong, note why in the write-up and set a better one for the next loop. Changing it while running is how a rule becomes a preference.
Does this replace a roadmap?
It sits under one. The roadmap says what you would build; the loop says whether you have earned the right to build it yet.
What if the week's question turns out to be the wrong question?
Write that as the answer. A week that establishes the question was wrong is a productive week, and it is the most common outcome in the first month.
