Search Engine Journal columnist Harry Clarkson-Bennett analyzed Similarweb traffic across 15 of the UK’s largest publishers and found that 18-to-34-year-olds hold 29.5 percent of the average audience, slightly above the UK’s 28 percent population benchmark from the Office for National Statistics. On share alone, publishers look roughly representative of the country they serve. That reading is wrong, and the mechanism explaining why is the real story.
Share has barely moved over three years. Premium publishers lost 3.0 percentage points of younger-audience share, public-service outlets lost 2.6 points, and popular publishers lost just 1.0 point. Platforms, the article’s term for social and video services, actually lost more share than any publisher segment: 6.8 percentage points.
Those numbers invite a comfortable conclusion: publishers are holding steady, and platforms are the ones losing young people. The comfort collapses once total audience size enters the picture. Total traffic across the 15 publishers fell 12 to 32 percent over the same period, and the under-35 slice fell faster than the total in every single segment.
In real terms, popular publishers lost 34.2 percent of their younger audience, premium publishers lost 30.7 percent, and public-service publishers lost 16.9 percent. Platforms lost only 9.2 percent by volume, the smallest decline of any segment, despite posting the largest percentage-point drop in share. A shrinking older audience was masking a far larger younger-audience exodus.
The comparison to platforms sharpens the point. Platforms average 49.2 percent under-35 audience, about 1.7 times what publishers average. No traditional publisher in the sample clears 40 percent; the New York Times leads the group at 39.1 percent for its UK audience, with the BBC next at 35.1 percent. Clarkson-Bennett’s data does not show those departing publisher readers landing on platform websites. Platforms’ own older audience grew more than 10 percent in real terms over the same window, so the destination of the missing under-35 traffic is not established by this dataset. The analysis is limited to website visits and excludes app usage, and Similarweb’s estimates carry the usual third-party measurement caveats.
The same blind spot applies to any SEO or content team that reports growth or decline only as a single blended year-over-year traffic number. A flat or even rising topline figure can conceal a cohort in freefall if a stable or growing segment offsets it, whether that segment is age, device, geography, or acquisition channel. Search Console and platform analytics default to aggregate views. The finding here only becomes visible once the audience is segmented and measured in absolute terms rather than share.
Any team that reports traffic as one blended trend line should rerun the number by cohort before the next quarterly review. Three years of stable topline share masked an under-35 audience declining by as much as 34.2 percent in volume, and the same masking could be hiding a comparable problem in any segmented view a team has stopped checking.
Search Engine Journal published Harry Clarkson-Bennett’s analysis on July 2, 2026.
