Dwell Time

Dwell time is the number of seconds a visitor actively engages with a specific section before scrolling away or leaving. Apple tracks it through cursor velocity, scroll pauses, and hover duration on elements. Their 2024 tests showed bento sections holding users 47 percent longer than the three column feature rows every SaaS site copied between 2018 and 2023. The reason is behavioral. Readers arrive with TikTok reflexes. They refuse to follow your forced linear order. They scan for the cell that matches their exact pain point right now. A bento grid respects that impulse with multiple entry points and clear size based priority. That respect turns into psychological investment. Investment stretches the seconds spent before they bounce.

What it is not is any of the vanity metrics most dashboards still worship. It is not time on page. Time on page happily inflates itself by counting the four minutes your prospect spent on a support call with your tab open in the background. Dwell time demands active signals. It is not overall session duration or bounce rate. Those numbers only tell you what happened. Dwell time diagnoses why it happened in the first 12 seconds. It is not a content length hack. Adding paragraphs to every cell does not buy more time. It triggers the exact opposite reaction. Readers see dense text walls and leave faster. The metric rewards compression, deliberate visual hierarchy, and the confidence to let some cells stay small.

Concrete examples make the lift obvious. Apple Mac pages in 2026 deliver the cleanest proof. The MacBook Pro M4 section uses one dominant 2x2 cell with the laptop floating on a dark field. Supporting cells hold single metric callouts for battery life, neural engine speed, and display performance. A vertical testimonial cell runs down the right edge. Real user data shows 29 seconds average dwell time. The 2022 three column version with identical icons and copy managed 17 seconds. Nothing changed but the layout. Linear repeated the win on their features page. The command bar takes the 2x2 hero slot with an embedded 4 second GIF of keyboard shortcuts firing. Four smaller cells use strict monospace labels for instant search, Git integration, and cycle time metrics. Engineering visitors average 24 seconds here and interact with at least three cells before clicking through. That engagement level never appeared in their old equal width feature row design.

Vercel mixed media types on their 2026 homepage without breaking coherence. One cell shows an AI agent workflow illustration. Another holds a live edge function log. A third contains a customer quote from Nike with logo. Every cell shares the same 16 pixel radius, 32 pixel internal padding, and dark background. The A B test delivered a 53 percent dwell time increase from 14 seconds to 22 seconds. Conversion from that section rose 34 percent. Figma applied the same logic to their FigJam 2 launch page in 2025. The AI brainstorming cell got the hero treatment. Supporting cells covered real time cursors, template library depth, and enterprise SSO. Dwell time hit 34 seconds and became their highest converting marketing asset that quarter. Stripe adopted a dashboard style bento for their billing page refresh. The revenue overview cell sat at 2x2 while smaller cells showed fraud protection stats, payout speed, and integration logos. Time on section climbed from 9 seconds to 18 seconds. Upgrade rate followed it up 22 percent. Notion saw similar gains on their template gallery. The 2024 redesign moved from symmetrical cards to intentional bento spans. Average dwell time jumped from 11 seconds to 19 seconds with no copy changes.

The teams that win at this metric share three habits. They size cells according to business priority not aesthetic balance. The hero capability always gets the biggest real estate. They limit every cell to one tight line of copy. Longer text kills momentum instantly. They enforce visual coherence with identical backgrounds, radii, and padding across every compartment. Break any of these rules and dwell time collapses. The pattern is not about looking current. It is editorial discipline translated into layout.

Track and optimize for dwell time when your product is a platform with one primary capability and six supporting ones of uneven weight. Use it on marketing pages that must prove breadth at a glance without forcing a reading order. Measure it on proof pages, integration roundups, and dashboard style sell sheets. Set up custom events that ping every four seconds of active mouse movement inside the viewport. Anything under 15 seconds on a primary features section means your layout is lying about how interesting the product actually is.

Stop chasing it on linear narratives. Step by step tutorials, onboarding sequences, and long form case studies die inside bento grids because readers lose the sequence and bail. Avoid it when every feature carries genuinely equal weight. In those rare cases the honest three column row still wins. Never optimize for dwell time on pricing tables, checkout flows, or support documentation. Those surfaces exist to remove friction and drive decisive action, not to encourage grazing. Context beats pattern every single time.

Dwell time is the only honest report card your marketing page receives.