Beyond SEO: New Technique Boosts Website Visibility in AI-Powered Search Engine Results

Researchers from Princeton University, Georgia Tech, the Allen Institute for AI and IIT Delhi have created a set of guidelines for optimising websites so they rank higher in AI-powered search engine results.

Large language models (LLMs) have revolutionised the world of search engines, which now use generative models to provide more accurate and personalised results for users.

Systems like BingChat, Google’s SGE, and perplexity.ai combine the strength of conventional search engines with the flexibility of generative models.

However, website and content creators are finding it hard to control how and when their content is used by these generative engines, risking disadvantage to the creator economy.

To help content creators ensure their work is seen by users, GEO (Generative Engine Optimization) has been introduced, offering a way to optimise content visibility in generative engine responses by benchmarking and enhancing visibility metrics.

GEO-bench, a new benchmark query suite, has been developed to provide content to respond to user queries across a variety of domains.

Evaluation has shown GEO boosts visibility in generative engine responses by up to 40%, highlighting the need for domain-specific methods.

The researchers tested nine different optimisation techniques on sites covering legal, historical, business and scientific topics.

Each technique performed better for some topics than for others, so it's best to tailor optimisation to the subject matter.

Three key strategies that worked across the board were citing sources, adding quotations and including statistics.

Reaserch Paper: GEO: Generative Engine Optimization