Deep dive: scroll depth lies — measure engaged time instead

Deep dive: scroll depth lies — measure engaged time instead

Scroll depth is the default "are they reading" metric, and it's quietly misleading. A visitor can scroll your entire page in two seconds and read none of it. The research community moved to a better proxy years ago, and most affiliate landers haven't followed.

Chartbeat's large-scale analysis (billions of pageviews) found a weak relationship between scrolling and actual reading — many users scroll well past content they never engaged with, a behavior they called the gap between "scrolling" and "reading." Their more predictive metric was engaged time: cumulative seconds where the user is demonstrably active (scrolling, moving, interacting), not just seconds the tab was open. Engaged time correlated with the things you actually care about — return visits, sharing, conversion — far better than scroll depth or raw time-on-page.

The mechanism behind the lie: scrolling is cheap and often reflexive. A bored user scrolls faster, hunting for a reason to stay or an exit, which inflates scroll depth precisely when engagement is collapsing. So a page with deep scroll and shallow engaged time isn't winning — it's being scanned by people looking for the door.

For practical measurement on a lander: instrument engaged-time buckets, not just scroll milestones. Where does active time concentrate, and where does it die? A section that everyone scrolls past but no one dwells on is dead weight; a section with high engaged time but low scroll-through is a bottleneck where attention is captured but momentum stalls.

The diagnostic value is sharp: deep scroll + low engaged time = skimming toward the exit; shallow scroll + high engaged time = people stuck reading something (good if it's your proof, bad if it's a confusing section). Don't optimize the metric that's easy to move; optimize the one that predicts the outcome.

TL;DR
— Chartbeat's data: scroll depth correlates weakly with actual reading; engaged time predicts conversion and sharing far better.
— Bored users scroll faster, inflating depth exactly when engagement collapses — deep scroll can mean hunting for the exit.
— Instrument engaged-time-by-section to find dead weight (scrolled-past, no dwell) and bottlenecks (high dwell, low scroll-through).