Google appears to be relying more heavily on structured signals such as schema.org and og:image to determine thumbnails in Search and Discover, making image optimization a growing visibility factor.
For years Google has often selected thumbnails for Search and Discover in ways that felt unpredictable. Publishers would carefully choose a featured image only to see Google display a completely different one in the results. Recently however several SEO observations suggest that Google may be relying more consistently on structured signals when deciding which images to surface as thumbnails.
Specifically elements such as schema.org image properties and og:image metadata appear to be playing a stronger role in how Google selects the primary visual representation of a page across Search features and Discover feeds. While Google has long supported these signals, the practical impact on visibility seems to be increasing as visual presentation becomes a more important part of how content is consumed in modern search interfaces.
This shift is not surprising. As Google integrates more AI-driven summarization layers and richer content previews, selecting the right image becomes critical for both user engagement and contextual understanding. Instead of randomly pulling visuals from within the page body, the system increasingly appears to prioritize explicit publisher signals that define the preferred representation of the content.
In practical terms this means that image optimization is no longer just about file size, alt attributes or basic SEO hygiene. Publishers now need to clearly communicate which image represents the page. If multiple strong images exist within an article and no explicit preference is declared, Google may still choose an unexpected thumbnail that does not align with the intended visual narrative.
Structured data therefore becomes a control mechanism. By specifying images in schema markup and aligning them with Open Graph signals such as og:image, site owners can reduce ambiguity and guide Google toward the preferred thumbnail for Search and Discover surfaces. This is particularly relevant for news publishers, blogs and editorial sites where thumbnails strongly influence click-through rates.
As Search continues to evolve toward richer visual interfaces and AI-generated previews, the role of structured signals in image selection will likely grow even further. Clear signals reduce guesswork for the algorithm and help ensure that the visual context presented to users matches the publisher’s intent.
In short the era of random thumbnails may be slowly fading. The more explicitly a page defines its preferred visual identity, the easier it becomes for Google to surface that representation consistently across search experiences.
