Event types

Learn about the different events you can collect with the Algolia Insights API and the features they unlock.

The Algolia Insights API accepts three types of events: click, conversion, and view events. By sending events, you can enable one or more Algolia features.

To get started with your implementation, see Get started with click and conversion events.

Events by features

The following table lists events and where they’re used.

	NeuralSearch, Dynamic Re-Ranking, Query Categorization, Search analytics, Revenue analytics, A/B testing	Recommend	Personalization
clickedObjectIDsAfterSearch
convertedObjectIDsAfterSearch
addedToCartObjectIDsAfterSearch
purchasedObjectIDsAfterSearch
clickedObjectIDs
convertedObjectIDs
addedToCartObjectIDs
purchasedObjectIDs
clickedFilters
convertedFilters
viewedObjectIDs
viewedFilters

To learn more about the properties of each event, see the Insights API reference.

Minimum number of events

In general, the features that require events work better with more events. Some features need a minimum number of events to be active.

Events requirements for NeuralSearch

• Minimum 1,000 click OR 100 conversion events within 30 days.

NeuralSearch uses these events:

• clickedObjectIDsAfterSearch (click)
• convertedObjectIDsAfterSearch (conversion)
• addedToCartObjectIDsAfterSearch (conversion)
• purchasedObjectIDsAfterSearch (conversion)

Events requirements for Recommend—Frequently Bought Together

• Minimum 1,000 conversion events with 2 or more products within 30 days
• Maximum 3,000,000 conversion events within 30 days

If there aren’t enough events within 30 days, the collection period extends to 90 days.

Frequently Bought Together uses these events:

• convertedObjectIDsAfterSearch
• convertedObjectIDs
• addedToCartObjectIDsAfterSearch
• addedToCartObjectIDs
• purchasedObjectIDsAfterSearch
• purchasedObjectIDs

Events requirements for Recommend—Related Products

• Minimum 10,000 events within 30 days
• Maximum 3,000,000 events within 30 days

If there aren’t enough events within 30 days, the collection period extends to 90 days.

Related Products uses these events:

• clickedObjectIDsAfterSearch
• clickedObjectIDs
• convertedObjectIDsAfterSearch
• convertedObjectIDs
• addedToCartObjectIDsAfterSearch
• addedToCartObjectIDs
• purchasedObjectIDsAfterSearch
• purchasedObjectIDs

Events requirements for Recommend—Related Content

• Minimum 10 events on items with content-based attributes within 30 days
• Maximum 1,500,000 events within 30 days

If there aren’t enough events within 30 days, the collection period extends to 90 days.

Related Content uses the same events as the Related Products model.

Events requirements for Recommend—Trending Items and Trending Facet Values

• Minimum 500 conversion events within 30 days
• Maximum 3,000,000 conversion events.

Trending Items and Trending Facet Values use these events:

• convertedObjectIDsAfterSearch
• convertedObjectIDs
• addedToCartObjectIDsAfterSearch
• addedToCartObjectIDs
• purchasedObjectIDsAfterSearch
• purchasedObjectIDs

Events requirements for Dynamic Re-Ranking

• Minimum 20 click events OR 2 conversion events within 30 days

Records with fewer events won’t be re-ranked.

Dynamic Re-Ranking uses these events:

• clickedObjectIDsAfterSearch (click)
• convertedObjectIDsAfterSearch (conversion)
• addedToCartObjectIDsAfterSearch (conversion)
• purchasedObjectIDsAfterSearch (conversion)

To prevent a manipulation of your ranking, Dynamic Re-Ranking only considers 1 event per user (userToken) per search query per record (objectID).

Events requirements for Query Categorization

Query Categorization needs a minimum number of events to train the model for predicting categories. It always uses events from the last 90 days.

Query Categorization uses these events:

• clickedObjectIDsAfterSearch
• convertedObjectIDsAfterSearch
• addedToCartObjectIDsAfterSearch
• purchasedObjectIDsAfterSearch

Events on category pages

You can build category pages with Algolia by sending a request with an empty search query and an applied filter.

Because category pages are just a different form of search requests, you’ll use the same events for category pages and search results.

Events related and unrelated to Algolia requests

Some events come in two variants: with and without a query ID. Events with a query ID have AfterSearch in their name.

Most Algolia features use events with query IDs, for example, to calculate the conversion rate of your search.

What is the query ID?

The queryID is a random sequence of letters and numbers that uniquely identifies a search request. Algolia uses the query ID to connect a search or browse request to a click or conversion event. Use the events debugger to check which search API request corresponds to which event. Algolia generates the query ID and includes it in the API response if you set clickAnalytics to true when making the search request.

Since conversions often occur outside Algolia-driven experiences—for example, from the shopping cart, you need to keep track of the query ID on your site.

Events related to Algolia requests

To track events after a previous Algolia search or browse request, set the clickAnalytics parameter to true and use the clickedObjectIDsAfterSearch or convertedObjectIDsAfterSearch methods.

Examples include:

• A user searches and clicks a search result to visit a detail page. You can track this by sending a clickedObjectIDsAfterSearch event.
• A user converts on a category page you built with Algolia. You can track this by sending a convertedObjectIDsAfterSearch event.

Click and conversion events with a queryID are time-sensitive. They must occur within one hour of the corresponding search request.

Events unrelated to Algolia requests

To track events that are unrelated to a previous Algolia search or browse request, use the clickedObjectIDs and convertedObjectIDs methods.

Examples include:

• You’re not using Algolia to build your category pages. Use clickedObjectIDs to track clicks and convertedObjectIDs to track conversions.

• You’re sending emails with recommendations. When users click a recommendation and buy an item, track the conversion with convertedObjectIDs.

• In a buy-online, pick-up in-store scenario, the conversion finishes hours or days after the initial search or browse request. Use convertedObjectIDs to track the conversion.