Google’s Gary Illyes and Martin Splitt addressed the rise in web page sizes on a recent podcast, arguing that heavier pages are not inherently bad because much of the added weight comes from structured data, metadata, and machine-facing content that serves legitimate purposes even if users never see it. The median web page grew from 845 kilobytes in 2015 to 2.3 megabytes in July 2025, but Splitt noted that compression, differing definitions of page size, and the ratio of content to markup all make raw page weight a misleading benchmark. Illyes dismissed the idea of separating human-facing and machine-facing content into separate documents as impractical, saying Google’s experience with mobile and desktop page splits showed that separate content versions consistently diverge, and that the approach would also amplify spam.
