Field notes
Notes from the table, not from a launch calendar
Short pieces on the judgments that show up in a feature adoption brief or a user-behavior reading: which week to trust, which columns to send, and when a cliff is only a second name for the same tap.
-
12 June 2026 · Niran Chaiyasit
The first week is a poor ruler for feature adoption
The week a feature ships mixes curiosity, staff checking, and the banner. It is a poor ruler for whether anyone came back.
-
3 July 2026 · Mira Adelman
What belongs in the event table before a behavior reading
A path reading stands or falls on four fields. Everything else in the export is either useful context or a risk you did not need to take.
-
14 August 2026 · Niran Chaiyasit
Marks that hold when the sample is small
A steep curve drawn from forty people is a costume. Counts, dots, and a stated floor keep a small sample honest.
-
2 September 2026 · Mira Adelman
When the drop-off is a label, not a habit
Some cliffs in a path chart are two names for one tap. Joining them is a drawing decision, and it should be visible on the plate.