Personalized recommendations

Personalized recommendations re-rank a model’s results using each user’s affinities. Two users who view the same product see the same model, but a different order of recommended items.

This is a beta feature according to Algolia’s Terms of Service (“Beta Services”).

Personalized recommendations are compatible with both Classic and Advanced Personalization. Personalized recommendations are available for the following models:

How personalized recommendations work

Personalized recommendations use the same Recommend models, but they incorporate user affinities to tailor the recommendations to individual users. The recommendations are re-ranked based on users’ affinities, which are derived from their interactions and preferences. Each Recommend request must include the userToken. Without it, Recommend has no affinities to re-rank with and returns the model’s default results.

Before you begin

This feature isn’t available on every plan. Refer to your pricing plan to see if it’s included.

Before you enable personalized recommendations:

Set up your Personalization configuration and confirm that user affinities are being computed.
Check that you send a userToken with your events, and that each user has their own token.

Enable personalized recommendations

Go to the Algolia dashboard and select your Algolia .
On the left sidebar, select Recommend.
Select one of the supported models and click Start using.
On the model’s configuration page, turn on Personalization.

Send a user token with your requests

Recommend inherits enablePersonalization from your Personalization configuration, so you only need to send it to override that setting for a single request. If you use InstantSearch with the insights middleware, the Recommend widgets send the user token for you. You can also set userToken in the widget’s queryParameters:

JavaScript

relatedProducts({
  container: "#relatedProducts",
  objectIDs: ["OBJECT_ID"],
  queryParameters: {
    userToken: "user-1234",
    enablePersonalization: true, // Optional: overrides the configuration
  },
});

If you use the Recommend API client, add userToken to the queryParameters of each request:

JavaScript

const response = await client.getRecommendations({
  requests: [
    {
      model: "related-products",
      indexName: "INDEX_NAME",
      objectID: "OBJECT_ID",
      threshold: 0,
      queryParameters: {
        userToken: "user-1234",
        enablePersonalization: true, // Optional: overrides the configuration
      },
    },
  ],
});

Send the same token you send with your click and conversion events. For anonymous users, get the token from Search Insights.

Personalize fallback results

When using a fallback query, the results aren’t personalized by default. To personalize fallback results, set enablePersonalization: true in fallbackParameters:

JavaScript

relatedProducts({
  // ...
  fallbackParameters: {
    // ...
    enablePersonalization: true,
  },
});

If you use the Recommend API client, set both userToken and enablePersonalization in fallbackParameters.

Search and Discovery platform

Optimization and Personalization

AI-powered experiences

Production and scale

Personalized recommendations

How personalized recommendations work

Before you begin

Enable personalized recommendations

Send a user token with your requests

Personalize fallback results

See also

​How personalized recommendations work

​Before you begin

​Enable personalized recommendations

​Send a user token with your requests

​Personalize fallback results

​See also

How personalized recommendations work

Before you begin

Enable personalized recommendations

Send a user token with your requests

Personalize fallback results

See also