Maya nearly reprioritises a quarter around three furious emails about the export flow, finds the complaint rate is under a tenth of a percent, and then finds that all three come from accounts carrying most of the revenue
Three Angry Emails — Volume I, Understanding Humans Before Products · Seeing Reality · Dilemma · Core. Maya nearly reprioritises a quarter around three furious emails about the export flow, finds the complaint rate is under a tenth of a percent, and then finds that all three come from accounts carrying most of the revenue
01 · Opening
Maya nearly reprioritises a quarter around three furious emails about the export flow, finds the complaint rate is under a tenth of a percent, and then finds that all three come from accounts carrying most of the revenue
02 · The setup
Three emails in one week, all furious, all about the new export flow. Maya reads them twice.
Maya thinks: Three in a week. Something is badly wrong.
03 · The setup
Sam: Three customers in seven days. I need this on the roadmap this week.
04 · The evidence
Dev: Three out of how many?
Maya: I don't actually know.
05 · The evidence
The complaints are real. So is the denominator nobody had looked up.
Export flow — last 30 days — Complaints received: 3; Accounts using export weekly: 4,182; Complaint rate: 0.07%; Export completion rate: 98.4%
06 · The evidence
Three is not a quantity until you know what it is three out of. Three from a hundred is an emergency. Three from four thousand is Tuesday. The number in your inbox is the numerator, and the numerator alone can be made to justify anything.
This is why complaints feel like trends. They arrive one at a time, in sequence, each one vivid — and the four thousand people for whom the flow worked send nothing at all.
07 · The evidence
Maya: Nought point nought seven percent. It's noise.
Sam: Tell that to the three.
08 · The evidence
Priya: You answered how many. You haven't answered who, or how bad.
09 · The evidence
Same three accounts, looked up a second way.
The three accounts — Plan: Enterprise; Share of total revenue: 31%; Export runs per week: 600+; Renewal date: within 90 days
10 · The evidence
How many, out of how many? — Get the rate before you get an opinion. This alone kills most false alarms, and it is one query.
How bad is it for the ones affected? — A rate cannot see severity. Data loss for eleven people outranks mild annoyance for eleven thousand.
Who are they, and are they load-bearing? — Concentration beats volume. A tiny percentage of accounts can carry most of the revenue, the reference calls, or the contract you cannot lose.
11 · The evidence
Both readings were right. It was noise by volume and urgent by concentration, and Maya had only asked one of the two questions.
Maya thinks: The rate told me to ignore it. The rate wasn't the whole question.
12 · You make the call
The story does not tell you first.
A. Deprioritise it — the rate proves it is not a real problem
B. The rate settles how widespread it is, not how much it matters
C. Ship the fix — three complaints in a week is a clear trend
13 · What happened
Maya: We fix it. Not because there were three, because of which three.
Sam: I'll take that to them.
14 · Complete
Get the denominator before the opinion — then ask how bad, and for whom. Volume and severity are two questions, and the loud ones are sometimes right.