The cleanup that dropped "sessions" 22% and made every metric truer
A content site's traffic looked great and their ad RPM looked terrible. The two numbers didn't fit. Something was inflating sessions without spending a cent.
The bot-cleanup playbook:
— Sort pages by sessions and flag any with 100% bounce and 0:00 duration
— Check Tech details: a flood from one browser version or screen resolution is a tell
— Pull the source list and look for unfamiliar referral domains repeating
— Build a segment excluding those signals and compare engaged sessions
— Apply server-side filtering for the worst offenders, don't rely on GA's checkbox alone
Real engaged sessions were 22% below the headline number. Once filtered, average session duration rose from 0:48 to 1:14 and RPM finally made sense.
Lesson: inflated traffic doesn't just flatter you, it hides your real engagement. Clean the data before you read it.
The Pixel Diary
@thepixeldiaries
The cleanup that dropped "sessions" 22% and made every metric truer
Этот пост опубликован в Telegram-канале The Pixel Diary. Подписаться можно по ссылке: @thepixeldiaries.