Scroll depth: how many visitors actually see the bottom of your landing page?

A landing page is designed top to bottom, but rarely read to the bottom. Between the first screen and the footer, every section loses a share of the audience — and arguments placed too low speak into the void. Scroll depth measures precisely that attrition: what percentage of visitors reaches 25%, 50%, 75%, 100% of the page. It's easy to collect, free in GA4, and yet missing from most dashboards — even though it answers the question every page owner asks: are my visitors even seeing what I have to say?

What scroll measures — and what it doesn't

Let's be precise up front: scrolling isn't reading. A visitor can sweep through the page in two seconds of mouse wheel without absorbing anything, or spend two minutes on the first screen. Scroll depth measures exposure — which sections had a chance to be seen — not attention, which is better read in time spent on the page and in heatmaps, whose scroll maps are in fact the visual rendering of the same data. The three metrics complement each other: scroll says how far, time says how much, the heatmap says exactly where. Survey methodology has studied scrolling closely: a study by Peytchev, Couper, McCabe, and Crawford published in 2006 in Public Opinion Quarterly ("Web Survey Design: Paging Versus Scrolling") compared, across more than 10,000 respondents, questionnaires presented as one long scrolling page versus successive short pages, measuring how the scrolling format changes completion behavior — scrolling isn't a neutral gesture, it's an engagement decision that design can encourage or discourage. The authors extended this work to mobile, where Mavletova and Couper (2014, "Mobile Web Survey Design: Scrolling Versus Paging") show the trade-offs differ on small screens — a useful reminder: your desktop and mobile scroll curves must be analyzed separately.

Measuring scroll depth in GA4

GA4 includes scrolling in its "enhanced measurement": the scroll event fires automatically when the visitor reaches 90% of the page. That's a start, but one threshold doesn't draw a curve. To get the intermediate steps, create a "scroll depth" trigger in Google Tag Manager at 25, 50, 75, and 90%, sent as a GA4 event with the percentage as a parameter. You can then read, page by page, the share of sessions reaching each step — and cross it with conversions, in the spirit of our guide to conversion tracking in GA4. Heatmap tools (Clarity, Hotjar) provide the same data visually, with the exact drop-off position in pixels rather than percentages.

Reading the curve: the three typical profiles

The early cliff: the first screen doesn't invite

If a large share of visitors never even passes 25%, the problem isn't the page's length: it's the first screen, which gives no reason to go down. Vague headline, missing promise, decorative visual — the diagnosis matches that of the fold: the top of the page must both deliver the essentials and prime what follows (a visually cut-off element peeking through, an arrow, the start of a visible section).

Steady erosion: normal, optimize at the margins

A gradual decline is the healthy profile: every section loses a little audience — that's the price of length. The decision becomes a question of order: are your most decisive arguments in the zone the majority still sees? If your strongest social proof sits at 80% depth where only the most motivated survive, move it up — the trade-off between primacy and recency is settled in light of this curve, and the long vs short page debate finds its empirical answer here, specific to your traffic.

The localized drop: one section drives people away

A sharp fall between two steps points to a guilty section: an indigestible wall of text, a premature technical table, a video that breaks the rhythm, or a false ending — a section that visually looks like a conclusion and makes people believe the page is over. It's the most profitable case to fix, because the solution is local: rewrite, lighten, or move a single section, then check the step has vanished from the curve.

Three concrete decisions scroll data enables

• Place the CTA where visitors actually are — if half the audience stops mid-page, a single final button works for the other half only; repeat the CTA at the depths still populated (see how many CTAs on a landing page).
• Cut the dead sections — a section almost nobody reaches that doesn't serve the conversion of the most motivated is maintenance cost with no benefit: remove or merge it, and the page gains speed and clarity.
• Prioritize A/B tests by exposure — testing a section seen by 15% of visitors mechanically caps the measurable effect; the scroll curve is one of the best impact criteria for prioritizing your tests.

In practice

Set up the GTM thresholds once, let them run two to four weeks depending on your traffic, and read the curve separately on mobile and desktop before any structural decision. On the ten LanderKit templates (€89 each, €229 for the bundle), pages are organized as self-contained modular sections, so reordering the structure in light of your scroll curve — moving social proof up, relocating the FAQ, repeating a CTA — means moving blocks around, without touching the design or breaking the page.