No, it augments it. We believe AI search should be AI enhanced search. For each query, Algolia AI Ranking will use your users actions to Re-rank a result list that has been returned and ranked a first time based on your relevance configuration.

Yes, using our Visual Editor or our Rules API, you can merchandise the search experience or execute on editorial strategies on top of the AI-powered search results.

The outcome may appear similar — both approaches lead to changes in product ranking — but there are crucial differences. Boost and bury rules require manual configuration by merchandising teams. These rules aren't intended to enhance your search engine's performance; rather, they're designed to override existing search results. For instance, to promote slow-moving inventory, you might implement boost rules to artificially elevate certain products in search rankings.

Boost and bury rules are effective when you have specific objectives that require manual ranking adjustments. However, if your aim is to maximize a specific goal, incorporating signals enables our algorithm to automatically refine your product rankings and boost the probability of generating conversions.

AI Ranking blends a variety of ranking signals to optimize for conversion and other business goals:

• Events (e.g., clicks, conversions)
• Business metrics and data stored within individual Algolia records (e.g. margin, offline revenue, availability, shipping, sponsored brands, promotions)
• Relevance scores (e.g., textual relevance via Algola’s tie-breaking algorithm)
• Textual signals (e.g., number of typos, distance between words)

With AI Ranking, we provide the machine learning model with a variety of signals, allowing it to optimize towards a goal. For example, the model may learn that the farther apart words are in a query, the less relevant the search results become. While it might not be immediately obvious why textual signals are as important as user interactions or business metrics, including them ensures a more balanced and effective ranking model.