Originally written in French. Translated by AI — the meaning has been preserved, not the prose.
Main idea
"Should we favour the customer or the supplier when a dispute pits the two sides of a marketplace against each other?" comes up in the roadmap review, then with sales on a strategic account, then with support on a refund, then in a specification, then in a ticket. Each occurrence presents itself as a one-off: that customer, that dispute, that opportunity.
The repetition, however, is the main piece of information. A question that resurfaces in different forums, with different participants, months apart, is not a string of badly resolved special cases: it is the symptom of a rule nobody has written at the level where it applies. As long as it is missing, every forum redoes the whole reasoning, from material only those present hold.
The signal is observable without an investigation: you only have to count the places where the same question has been asked. A single occurrence says nothing; three occurrences in three contexts say the subject is not a case, it is a frame.
Why it matters
This supplies an admission criterion for a product decision register that depends neither on the size of the subject nor on its urgency: you document what comes back, not what makes noise.
It also changes the nature of the expected answer. Faced with a recurring question, producing yet another local decision is wasted work: what has to be produced is the line that will make the next local decisions derivable without a meeting.
Nuances and limits
Recurrence can come from something other than an absent rule: from a rule that exists but cannot be found, from a rule that is known but never held to, or simply from a context that changes fast enough to legitimately reopen the subject each time.
And not every recurring question reduces to a rule: some tensions — arbitrating between two customer segments of equal importance — are honestly replayed at each occurrence, because the balance depends on parameters that move.
Open questions
- After how many occurrences, and over what period, does a question deserve to be requalified as a rule to be written rather than as a series of cases?