Time in app rises 40 percent after a redesign and the team celebrates, until Maya finds the rise is concentrated on the screen where a one-click action became a three-click hunt
Forty Percent More Time — Volume I, Understanding Humans Before Products · Attention · Investigation · Core. Time in app rises 40 percent after a redesign and the team celebrates, until Maya finds the rise is concentrated on the screen where a one-click action became a three-click hunt
01 · Opening
Time in app rises 40 percent after a redesign and the team celebrates, until Maya finds the rise is concentrated on the screen where a one-click action became a three-click hunt
02 · The setup
The redesign has been live a month and the headline number is the best anyone has seen.
Sam: Forty percent more time in app. Engagement is up and I am telling everyone.
03 · The setup
Session duration, month over month — Before redesign: 6m 20s; After redesign: 8m 52s; Change: up 40%; Sessions per user: unchanged
04 · The evidence
Maya: Sessions per user did not move. They are not coming back more, they are staying longer.
Dev: Which is good if they are enjoying it.
05 · The evidence
Maya splits the extra two and a half minutes by where it is being spent.
Maya thinks: A number this big is never spread evenly. It is sitting on one screen.
06 · The evidence
Where the extra time went — Reading dashboards: up 4s; Composing reports: up 11s; Finding the export button: up 2m 17s; Exports completed: down 22%
07 · The evidence
Maya: Nearly all of it is people looking for a button that used to be on the first screen.
Sam: And I have been calling that engagement for a month.
08 · The evidence
Priya: It is under Reports now. I get there eventually. It used to take one click.
09 · The evidence
Two screens, both showing more time. On one of them that is the product working and on the other it is the product failing.
Maya thinks: The number cannot tell me which. It never could.
10 · The evidence
Pair it with an outcome on the same screen — Time on a reading screen goes with articles finished. Time on an action screen goes with the action completed. Duration alone is uninterpretable in both cases.
Ask which direction is good here — Longer is better on a screen built for reading and worse on a screen built for one click. Write down the intended direction before you look, or you will accept whichever way it moved.
Split before you celebrate — A headline duration averages every screen together. A 40 percent rise sitting entirely on one screen is a different fact from the same rise spread evenly, and only the split distinguishes them.
11 · You make the call
The story does not tell you first.
A. It was measured over too short a period
B. Longer means success on some screens and failure on others
C. Time in app is a vanity metric and should not be tracked
12 · What happened
Export returns to the first screen. Time in app falls back to six and a half minutes and exports recover, and nobody reports the fall as bad news.
Maya thinks: The best week we have had, and the engagement number went down.
13 · Complete
A duration is uninterpretable until it is paired with the outcome its screen exists to produce. Next: the feature people refuse to switch on because of what switching it on says about them.