Why most programmes stop after three weeks
The pattern is consistent enough to plan around. Week one produces enthusiasm and a wide spreadsheet. Week two produces a longer spreadsheet. Week three produces a document nobody has time to read, and by week four the exercise belongs to nobody.
There are two causes and both are structural rather than personal. The first is no scope limit, so every week is a slightly larger version of the last. The second is no named reader, so the output has no standard to meet and no one to complain when it slips. A framework is not an attempt at completeness; it is an attempt at something that still happens in month six.
Step one: name the list, and cap it
Five to twelve companies, in three groups, with the reason each one is there written beside its name and a date to review that reason.
- Direct: you lose deals to them, by name, and can point at one.
- Adjacent: they solve the same problem from a different starting point, and a buyer may choose them instead of choosing anybody in your category.
- Reference: they set what buyers expect of a product like yours without competing for the same budget. Usually the largest company your buyers already use.
Step two: fix the evidence types before you read anything
Decide in advance which material you will read every week. The point is not thoroughness, it is that a quiet week becomes a finding rather than a reason to go looking somewhere new. A type you cannot read every single week does not belong on this list.
- Each company’s careers board, for roles opened and roles withdrawn.
- Each company’s own pricing and packaging pages, compared against last week’s dated capture.
- Statutory registers and regulatory filings, where the sector has them.
- The company’s own announcements, read as statements of intent rather than as events.
- Independent reporting that adds something an announcement did not say.
- Customer-visible product documentation and release notes, which move before marketing does.
Step three: the rule that decides what counts
One written rule, three tests, applied to every item before it reaches the brief. This is the step that separates a framework from a reading habit, and it is the step most often skipped.
- Who said it: the company itself, a statutory register, or a party with no stake in the claim. Each carries different weight and the brief should say which it was.
- When they said it: a date carried by the item, not the date you read it. An undated item is not a finding yet.
- How many independent parties carry it: eight outlets repeating one announcement are one party, not eight. Merge the reprints before counting.
Leads and findings are different things
An item that passes the first two tests and fails the third is a lead. It goes in a log as a lead, with its date, and it is not written up as a finding. If a second independent party carries it next week it graduates; if nobody else ever does, it stays where it is and costs nothing.
Keeping that distinction in writing is what lets a brief be short without being thin. A reader who trusts that a finding means corroborated will read a one-page brief closely. A reader who has learned that half of it is speculation reads none of it.
Step four: the brief, and the length limit
One page, three headings, and a hard limit that is the whole discipline: what changed, what it might mean, what we are doing about it. Each item gets one sentence, its date, and a link to the item itself so the reader can disagree with the evidence rather than with you.
The limit forces a ranking, and the ranking is the intelligence. A nine-page digest is a transcript of the week and transfers the ranking back to the reader, who has less context than you and less time.
Step five: the decision log
Record the decisions the brief changed, with dates and in one line each. Three months in, this is the only thing that answers the question that decides whether the programme keeps its budget: what did this change?
It earns its keep in the other direction too. It records the weeks where nothing changed, which is how you learn that a competitor you read closely never actually moves a decision. That company belongs on a quarterly list, and taking it off the weekly one is a gain rather than a retreat.
The ninety minutes, allocated
Rough but worth writing down, because an unallocated hour becomes three.
- Fifty minutes reading and discarding. Most of the week’s material fails the rule, and that is the ordinary case rather than a bad week.
- Twenty minutes judging and corroborating the three or four items that survived.
- Fifteen minutes writing the page.
- Five minutes circulating it and updating the log.
Where software belongs, and where it does not
Step two and most of step three are mechanical, repetitive and get worse when a person does them under time pressure at the end of a Friday. Reading the same material on a schedule, dating each item, merging reprints and discarding what fails the rule is exactly the work to hand over.
Steps one, four and five are not. Which companies matter, what a change means for your own plan, and what you decided as a result are judgements that depend on things no tool is told. A product offering to make those for you is offering to guess, and the guess will be confident. Automate the reading and keep the thinking.