Multiple content creators across YouTube report viewership declines beginning in mid-August, with data suggesting platform-wide algorithmic changes may be responsible. Gaming channel Bellular News has documented substantial shifts in both viewing patterns and device usage that coincide with the August 13 implementation of YouTube’s new AI-based age estimation system.

Desktop Traffic Plummets, Mobile Viewing Rises

According to Bellular’s analysis, prior to August 13, desktop viewing matched or exceeded mobile viewing, with computer traffic outperforming mobile nine out of 12 days leading up to the date. Following August 13, desktop traffic fell sharply, never again surpassing mobile in the dataset.

The desktop-to-mobile viewing ratio flipped from 56% favoring desktop in July to just 39.3% after mid-August  – a 16.7% relative swing driven primarily by decreased computer views rather than increased mobile activity.

Engagement Ratios

Several creators, including Brodie Robertson and Linus Tech Tips, have also identified unusual patterns in engagement metrics. While absolute view counts decreased, like counts remained relatively stable, creating a shift in view-to-engagement ratios.

Bellular’s data shows their average views-per-like ratio changed from 26.4 before mid-August to 15.36 afterward, suggesting core viewers continue engaging with content while overall view counts decline. This pattern appears consistent across multiple channels and content types.

Bellular’s testing has also exposed potential issues with YouTube’s restricted mode functionality. Of 30 videos sampled by Bellular, 14 disappear when restricted mode is enabled, despite showing no restriction labels in YouTube Studio. These restricted videos average 250,000 views compared to 421,000 for unrestricted content, a substantial performance gap confirmed through multiple statistical analyses.

Revenue and Sponsorship Implications

Despite declining viewership, revenue per mille (RPM) metrics increase across affected channels, indicating that monetizable views remain relatively stable while overall counted views decrease. This suggests potential targeting of non-monetizable traffic, such as ad-blocked viewing or data scraping activities.

The findings raise concerns for sponsorship relationships, as many brand deals rely on CPM (cost per thousand views) metrics directly tied to view counts. Creators including Bellular caution that uncertainty around view accuracy could weaken sponsor confidence in these agreements, since many brand deals rely on CPM metrics.

YouTube notes that restricted mode has existed since 2010. However, the company has not addressed specific creator concerns regarding the shifts in viewership, device usage patterns, or engagement ratio anomalies documented by multiple channels.

Checkout Our Latest Podcast