Cutting-edge search interfaces such as Google Lens and chat-style engines built on mainstream large language models are changing the rules of discovery. The familiar keyword-and-backlink era is receding. Instead, users increasingly receive direct answers from generative search tools that synthesize content and often cite sources, which reduces clicks to the open web and forces marketers to rethink where and how they surface information.
The impact is measurable. Early studies point to roughly 60% of searches now resulting in no click to external sites, a trend driven by features like AI overviews. Visual search is growing too: one survey found about 10% of u.s. adults use visual search frequently, while Google reports nearly 20 billion visual searches per month on Google Lens. These shifts mean search engine optimization can no longer be solely about keyword density and backlinks; it must account for how generative engines select and present ´credible´ sources and how images and video are interpreted by machine perception.
Data remains central, but the shape of the data changes. Teams must collect richer image metadata, optimize alt text for semantic context, and consider image composition and video segmentation as discoverability signals. Research suggests generative search favors content with authoritative sourcing and clear citations, so content should be structured to make provenance explicit. To compete, retailers need new tooling that goes beyond classic ranking trackers: tools that analyze image composition, evaluate detailed metadata, rank video segments for searchability, compare visual performance across platforms, surface actionable recommendations, and spot emerging trends in visual queries.
Because generative search engine optimization or GEO is still evolving, marketers must adopt an experimental mindset. Practical first steps include A/B testing visual assets, varying content structures and citation patterns, and monitoring changes in user behavior as visual and generative features are adopted. Companies that move quickly to learn, test, and optimize content and data pipelines for visual and generative search will be better positioned to win in the next wave of online discovery. The article is written by Gediminas Rickevičius, senior vp global partnerships at Oxylabs, who advises a data-first, iterative approach to mastering GEO.
