Volume I · Trust · Lesson 1

It Says Two Minutes

The export says about two minutes. For the biggest accounts it is closer to forty, and there is one quarter to fix it.

Lesson 015 minFoundations

Judgment0 judgment

←→ to move through the storySwipe or use the arrows to move through the story

01 / 13

Read this lesson as text

The whole story in one page — 13 frames, about 5 minutes.

It Says Two Minutes — Volume I, Understanding Humans Before Products · Trust · Dilemma · Foundations. An export promises two minutes and sometimes takes forty, and Maya has to choose between making it faster and making it honest

  1. 01 · Opening

    It Says Two Minutes

    The export says about two minutes. For the biggest accounts it is closer to forty, and there is one quarter to fix it.

  2. 02 · The setup

    Dev brings it up the way people bring up things they have mentioned before.

    Dev: The export dialog says about two minutes. For the big accounts it is closer to forty.

  3. 03 · The setup

    Export duration, last 30 days — Median: 1 min 50s; 90th percentile: 9 min; 99th percentile: 38 min; Dialog says: about 2 minutes

  4. 04 · The evidence

    Maya: The median is honest. It is the tail that is lying.

    Dev: The tail is our largest accounts. It is the only tail that matters.

  5. 05 · The evidence

    The support notes are all the same shape. Nobody complains about waiting. They complain about being told it would be over.

    Maya thinks: They stayed at their desks for two minutes and then had to decide, over and over, whether it had broken.

  6. 06 · The evidence

    Sam: Two accounts asked me last week whether the export is reliable. It is reliable. It just does not look it.

  7. 07 · The evidence

    Two options, one quarter, and Dev is clear that only one of them is actually available.

    Dev: I can maybe halve the tail. Forty becomes twenty and the dialog is still wrong.

  8. 08 · The evidence

    Maya: Or we spend the same weeks telling them exactly where it is and stop promising a number we cannot keep.

  9. 09 · You make the call

    Dev can maybe halve the 38-minute tail this quarter, or the same weeks can make the export screen honest. Which fixes what customers are actually complaining about?

    The story does not tell you first.

    A. Halve the tail, because the wait is the problem

    B. Keep the two-minute estimate and add a progress bar

    C. Make it honest, because the broken promise is the problem

  10. 10 · What happened

    They ship the honest version first, because it is testable in a fortnight and the other is not.

    Export screen, after — Shows: rows completed of total; Shows: account size band; Promises: nothing; Median duration: unchanged

  11. 11 · What happened

    Trust is a promise kept, not a wait avoided

    A customer forgives a slow operation. What they do not forgive is being told it would be over and finding it was not, because that teaches them the product does not know what it is doing.

    Every unkept small promise spends a little of the credit you need for the large moments — an outage, a price change, a migration. Speed buys satisfaction. Honesty buys the benefit of the doubt, and only one of those is there when something actually goes wrong.

    This is why the honest version is the ambitious choice rather than the resigned one. Halving the tail still leaves the dialog wrong. Telling the truth makes it right at every duration, including the ones you have not measured yet.

  12. 12 · What happened

    The tail work happens the following quarter and lands quietly, because by then nobody is watching the clock in fear.

    Maya thinks: Same forty minutes. Completely different product.

  13. 13 · Complete

    It Says Two Minutes

    Speed buys satisfaction and honesty buys the benefit of the doubt, and only one of those is available when something goes badly wrong. Next: what an apology is worth on the third occurrence.