Ad layout rebuild for viewability and session revenue

Almost every viewability brief we receive turns out to be a density brief. The page carries eleven slots, four of them sit below the point where ninety percent of sessions have already ended, and removing those four raises revenue per session while the reported view rate improves on its own. The MRC threshold is simple and unforgiving, fifty percent of pixels in view for one continuous second for display, two seconds for video, and a slot that cannot clear it is not underperforming inventory. It is not inventory. How we decide a slot lives or dies: Scroll depth distribution against slot position, per template and per device, taken from your analytics rather than an industry benchmark. Then Active View rate and revenue per session, never revenue per impression, because the per-impression metric is precisely the one that makes a dead slot look profitable by ignoring the readers who left. A slot survives if it clears roughly fifty-five percent viewability on mobile and pays more than the slot above it loses by being pushed down. In-content slots are injected by paragraph and word count with a minimum distance rule, not at a fixed index, because a four-hundred-word how-to and a two-thousand-word ingredient explainer are not the same page and should not receive the same treatment. The rules we hold to: Refresh no faster than thirty seconds, only on viewable time, and never on a slot that has not yet been seen. Every container gets a reserved box before the request goes out, aspect-ratio or explicit min-height matched to the size map, because a layout that reflows on render trades a cent of yield for a reader. GPT lazy loading is tuned as three separate numbers rather than one, and the fetch margin comes from measured scroll velocity on your site, since the defaults are wrong in both directions depending on how fast your readers move. Density stays inside the Coalition for Better Ads thresholds, which is a floor on behaviour and not a target to approach. Beauty, galleries and commerce modules: On cosmetics templates the real contest is between display and affiliate for the same position. We settle it with money: measured revenue per thousand sessions for the product module in that slot against the display unit it would replace, run as a split. Usually the product module wins above the fold and display wins below it, but not always, and the answer moves with the season and with what your commerce team has negotiated that quarter. We will not build a paginated slideshow whose pagination exists to multiply ad calls. How we prove it: Every layout change ships as a split test. Readers are bucketed by hashed identifier into a key-value pair in Ad Manager so a person stays in one arm across sessions, both arms run at least fourteen days, and we report the change in revenue per thousand sessions with an interval around it. If the interval crosses zero, the change is reverted and the report says so. We would rather hand you three reverted tests and one that held than a redesign nobody can attribute to anything. We will not: Add an interstitial to a site whose readers arrive from search carrying a question. Run refresh under the policy minimum. Put a sticky footer over a mobile checkout. Add slot count to reach a revenue number for one quarter, which is the version of this work that gets sold most often and is the reason your view rate is where it is today. Wrong fit: Sites where we cannot touch template code and must work entirely through a tag manager. Publishers who want a competitor's layout reproduced, since their scroll distribution is not yours and the copy is exactly why the numbers never follow. Anyone who needs an answer inside a week. The measurement is the deliverable, and it takes the time it takes.