Engagement Signals Beyond Session Duration
Average session duration alone misleads. Here's the portfolio of engagement signals — scroll depth, interaction depth, engaged sessions, return frequency — that actually predicts conversion and retention.
Engagement Signals Beyond Session Duration
A portfolio of engagement metrics — scroll depth, interactions, engaged sessions, return frequency — that predicts conversion better than session duration alone.
Engagement signals beyond session duration are the set of behavioural metrics you combine to judge whether a visitor is actually engaging with your store, rather than just sitting on a tab. Average session duration is easy to game and easy to misread: a five-minute session can mean a confused shopper or an abandoned browser window.
A mature engagement model blends depth (how far down the page), breadth (how many pages and interactions), quality (engaged sessions, not just any session) and loyalty (return frequency). Together these signals tell you whether your traffic is converting-grade — and they make ASD one input among many rather than the headline KPI.
If your weekly report leads with average session duration, you are reporting on a single noisy proxy. ASD averages bounced visitors, idle tabs, and real shoppers into one number that rarely moves with revenue.
The fix is not a better duration metric. It is a portfolio: four or five complementary signals you read together, the same way a performance manager reads ROAS alongside new-customer rate and contribution margin rather than in isolation.
The four signals that replace ASD as the headline
Start with depth. Scroll depth tells you whether visitors actually reach your value props, reviews, and add-to-cart on a PDP. On most Shopify product pages, the add-to-cart button sits around 40-60% scroll on mobile — if 70% of sessions never get there, no copy test will save you.
Then layer interaction depth and engaged sessions. Pages per session captures lateral exploration (collection → PDP → reviews → cart). GA4 engaged sessions filters out the dead weight: a session counts only if it lasts 10+ seconds, fires a conversion, or has 2+ pageviews. Return frequency closes the loop — visitors who come back within 7 days convert at 2-4× the rate of first-touch traffic on apparel and beauty stores.
How to weight the portfolio against conversion
The weights are not universal. For a £45 AOV beauty SKU bought on impulse, scroll depth and engaged session rate matter most — the decision happens on one page. For a £400 considered purchase like a coat or an espresso machine, return frequency and pages per session carry more predictive weight because the shopper researches across sessions.
Run a simple correlation against converters versus non-converters over the last 90 days. The signals that separate the two groups by the widest margin are the ones to weight up. We typically see scroll depth + engaged session rate explaining 60-70% of the conversion-likelihood variance on PDPs, with ASD adding under 5% of incremental signal once those are in the model.
ASD without depth is a vanity metric
A session that lasts 4 minutes but never scrolls past the hero image is a confused shopper, not an engaged one. Always read average session duration alongside scroll depth and interaction count — never as a standalone KPI in a board deck.
Operationalising the framework in your stack
Most teams already have the raw signals — they're just scattered. Scroll depth lives in Hotjar, engaged sessions in GA4, return frequency in Klaviyo, pages-per-session in GA4 again. The work is consolidating them into one weekly engagement score per landing page and per traffic source.
Once unified, engagement signals become the input layer for funnel analytics: which step loses engagement first, which traffic source brings genuinely engaged visitors, which PDP template earns scroll past the fold. That's the upgrade — from reporting a vanity number to diagnosing where attention actually breaks.
Correlation with purchase probability by engagement signal (PDP traffic, apparel)
Frequently asked questions
ASD averages every session — including idle tabs, bounced visits, and accidental opens — into one number. A user who leaves a tab open while making coffee inflates ASD without engaging at all. Read it only alongside scroll depth and engaged session rate.
A standard session counts any visit. A GA4 engaged session requires 10+ seconds of active time, a conversion event, or 2+ pageviews. Engaged session rate is a much cleaner quality signal than total sessions or ASD.
GA4 ships a basic 90% scroll event out of the box, but you'll want richer thresholds (25/50/75/90%). Most heatmap tools (Hotjar, Microsoft Clarity) capture it automatically, and Metricuno's snippet records granular scroll depth per template without theme edits.
On PDP traffic, scroll-depth ≥75% and engaged-session rate consistently top the list (r ≈ 0.55-0.65 with purchase). Return frequency within 7 days is the strongest signal for considered purchases above €200 AOV.
Pages per session is one signal inside the portfolio. It captures lateral exploration but misses depth on a single page — a visitor who scrolls and reads one long PDP is more engaged than someone bouncing across five thin collection pages.
Yes, but de-emphasise it. Lead with engaged session rate and scroll depth, then show ASD as supporting context. If ASD moves but engaged session rate doesn't, the duration change is noise, not a quality improvement.
Each funnel step gets an engagement score (e.g. scroll-to-add-to-cart on PDP, interaction rate on cart). When a step's conversion rate drops, you look at which engagement signal dropped first — that's the diagnostic, not the conversion rate itself.
They can inflate sessions and pageviews but rarely scroll or interact authentically. That's another reason scroll depth and engaged session rate are more trustworthy than ASD or raw session counts — bots fail the engagement filter.
Aim for 4 weeks of stable traffic minimum, or 5,000 sessions per page template. Engagement signals are less noisy than conversion rate, so you get directional reads faster — but seasonal and campaign effects still distort short windows.
Yes — and they outperform generic site-visitor audiences by a wide margin. Try "scrolled 75%+ on PDP but did not add to cart" or "returned within 7 days without purchasing". Both convert 2-3× better than broad pixel audiences on Meta and Google.
Get an AI expert review of your site
Paste your URL — Metricuno's AI runs the same heuristic checks a senior CRO consultant would, scoring your page and prioritising the fixes that'll move conversion fastest.