Our users read our release notes backwards.

We analyzed scroll and read patterns on our changelog across 340,000 sessions over 12 months.

Read from top (newest first): 34%
Scrolled to bottom first, then read up: 41%
Jumped to a specific date anchor: 18%
Bounced without reading: 7%

The largest group started at the oldest entry and read forward. They were not catching up on what was new. They were reconstructing a history — trying to understand how the product had changed since they last paid attention.

We had been writing release notes for people who never left. Our actual readers were people who had been away and were trying to rebuild their mental model.

We added a "what changed since you were last here" summary at the top, personalized to the user's last login. Read time on the changelog dropped 44%, but feature adoption from changelog readers rose 31%, because they stopped reading history and started finding their missing features.

Are you writing your updates for the users who never left or the ones who just came back?