bounce rate punished every reader who landed on your page, read the whole thing, and left happy. it calls your best pages your worst.
i'm om yaduvanshi, 18, an applied ai developer building cloudline, a cookieless analytics tool. day 19 of 90, building in public.
the metric that lies
bounce rate counts a session as a bounce when it has one pageview. no second page. that's it. someone who read your entire post and closed the tab counts exactly the same as someone who landed, hated it, and left in two seconds.
docs pages, blog posts, landing pages, any page that answers a question. the pages that did their job get the worst scores. i spent years optimizing a number that punished good pages. never again.
what i measure instead
engaged time, paired with scroll depth. 25, 50, 75, 100 percent, per page.
a page with three minutes of engaged time and 80 percent scroll depth is a page that worked. a page with ten seconds and a 90 percent bounce rate could be either. that's the number i check first in the morning now, not the one i used to check.
the part i stay honest about
cloudline stores no IPs, sets no cookies, doesn't read user agents. a pageview is page, referrer, screen size, time. so there's no session stitching, no "this person came back thursday". engaged time is page-level, and same-visitor-twice counts twice. the trade is written down, not hidden.
and revenue attribution is on the roadmap, not in the dashboard. said plainly because it matters.
free while in early access. no card.
— om
Top comments (0)