Guides / Personalization / AI Personalization (beta) / Monitor personalization

Error codes

On this page

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

AI Personalization relies on valid, relevant, and well-structured Insights events to generate accurate user profiles. When you configure AI Personalization, your Insights events and search queries are checked to ensure they meet specific requirements. If they don’t, a combination of the following errors and warnings will be returned.

You must resolve any errors before continuing to configure AI Personalization.

Index errors

The following errors will be returned if the indices you want to enable AI Personalization on don’t meet the requirements.

Missing index

code: no_index

An index in your configuration doesn’t exist in your app. Ensure that the name of the index is spelled correctly or configure AI Personalization in the dashboard so you only select indices that exist.

Missing records

code: no_records

An index in your configuration doesn’t contain any records. Review your indices to ensure you are configuring the correct index. If the index you intend to use contains no records, send data to the index.

Events errors

The following errors will be returned if the events for the indices you want to enable AI Personalization on don’t meet the requirements.

Use the Events Debugger to monitor and debug the 3,000 most recent events.

AI Personalization only checks valid Insights events with objectIDs that were received by Algolia in the past 24 hours. Events with filters are not used.

When a primary or replica index is selected in the configuration, events for the primary index and its replicas are combined and then validated altogether.

Invalid events mapping

code: invalid_events_mapping

Some events in the eventsMapping don’t exist. Ensure the names and types of the events are spelled correctly or use the dashboard to configure AI Personalization.

Mismatched object IDs

code: mismatched_object_ids

Less than 80% of events for this index and its replicas, that have an objectID, correspond to a record in the primary index with a matching objectID. Review the index and ensure you aren’t sending events for objects that don’t exist in the primary index, were recently deleted from the index, or exist in another primary index.

Missing events

code: no_events

No events exist for this index or its replicas. Ensure you are capturing all important events related to your search and discovery experience.

Missing user tokens

code: no_user_tokens

None of the events in this index or replicas have a userToken. Ensure you are sending a valid userToken for all events.

Unique user tokens

code: single_user_token

All the events for this index or replicas have the same userToken. Ensure you are sending a unique userToken for each user.

Invalid user tokens

code: undefined_user_tokens

All the events for this index or replicas have a userToken of undefined, anonymous, or ''. Ensure you are sending a unique userToken for each user.

Missing object IDs

code: no_object_ids

Some of the received events for this index and its replicas don’t have objectIDs. Ensure you are sending events with objectIDs.

Low number of returning users

code: low_returning_users

Less than 1% of user tokens appear in events on consecutive days. If this is unexpected, check that you’ve implemented persistent user tokens. If this matches your expectations, implement a strategy to get more users regularly returning to your site.

Events warnings

To ensure the most accurate user profiles, avoid the following warnings related to your events. Fixing them isn’t mandatory but can be helpful.

Missing user tokens

code: some_events_with_no_user_tokens

Some, but not all, of the events for this index or replicas have no userToken. Ensure you are sending a valid userToken for every event.

Invalid user tokens

code: some_events_with_undefined_user_tokens

Some, but not all, of the events for this index or replicas have a userToken of undefined, anonymous, or ''. Ensure you are sending a unique userToken for every user.

Low number of events with object IDs

code: low_object_ids

Some, but not all, of the events for this index or replicas don’t contain objectIDs. Ensure you are sending events with objectIDs.

No event subtypes

code: no_event_subtypes

Received conversion events lack either addToCart or purchase subtypes. Ensure you are sending valid subtypes for all conversion events.

Low number of returning users

code: low_returning_users

Less than 10% of user tokens appear in events on consecutive days. If this is unexpected, check that you’ve implemented persistent user tokens. If this matches your expectations, implement a strategy to get more users regularly returning to your site.

Query errors

The following errors will be returned when the requirements for search queries haven’t been met.

AI Personalization only checks search queries that were received by Algolia in the past 24 hours.

When a primary or replica index is selected in the configuration, queries for the primary index and its replicas are combined and validated altogether.

No users

code: no_users

In the past 24 hours, none of your users have searched your personalized index. Ensure you’ve correctly configured the index for personalization.

No queries

code: no_queries

In the past 24 hours, none of your search queries targeted your personalized index. Ensure you’ve correctly configured the index for personalization.

Missing queries with personalization enabled

code: no_queries_perso_enabled

None of your search queries have personalization enabled:

Missing users with personalization enabled

code: no_users_perso_enabled

Search personalization is not enabled for any users. This check is similar to missing queries with personalization enabled, but verifies the unique user tokens received by the search API.

Missing queries with profile found

code: no_queries_profile_found

None of your search queries are associated with a user profile. Ensure that you are sending a valid userToken, users are assigned a persistent user token across sessions and that user tokens sent to the search API match user tokens from Insights events.

Missing users with personalized queries

code: no_users_profile_found

None of your users profiles are found for search queries. This check is similar to missing queries with profile found, but verifies the unique user tokens received by the search API.

Missing queries with personalization applied

code: no_queries_profile_applied

None of your search queries have been re-ranked by personalization. Ensure the attributes in your index match the affinities in the user profiles, so records that match the profile can be boosted.

Missing users with personalization applied

code: no_users_profile_applied

None of your users had search queries with records boosted by personalization. This check is similar to missing queries with personalization applied, but verifies the unique user tokens received by the search API.

Queries warnings

To ensure the highest impact of personalization on search results, resolve the following warnings related to your search queries. Fixing them isn’t mandatory but can be helpful.

Low number of queries with personalization enabled

code: low_queries_perso_enabled

Only a subset of your queries have personalization enabled:

Low number of users with personalization enabled

code: low_users_perso_enabled

Only a subset of your users have search personalization enabled. This check is similar to low number of queries with personalization enabled, but verifies the unique user tokens received by the search API.

Low number of queries with personalization found

code: low_queries_profile_found

Only a subset of your queries are associated with a user profile. Ensure that you are sending a valid userToken, users are assigned a persistent user token across sessions and that user tokens sent to the search API match user tokens from Insights events.

Low number of users with personalization found

code: low_users_profile_found

Only a subset of your users profiles are found for search queries. This check is similar to low number of queries with personalization found, but verifies the unique user tokens received by the search API.

Low number of queries with personalization applied

code: low_queries_profile_applied

Only a subset of your queries can be re-ranked by personalization. Ensure that user profiles have more than one or two affinities, so they can be matched with records attributes to boost them in search results.

Low number of users with personalization applied

code: low_users_profile_applied

Only a subset of your users had search queries with records that can be boosted by personalization. This check is similar to low number of queries with personalization applied, but verifies the unique user tokens received by the search API.

Attribute errors

Ignored regex

code: ignored_regex

This attribute uses a reserved name such as objectID. Ensure that none of the attributes in your index uses a reserved name.

No values

code: no_values

You don’t have any values for this attribute in your index. Ensure that every record in your index has a value for this attribute.

All invalid values

code: all_invalid_values

All records in your index have an invalid value for this attribute. Common examples of invalid values include unknown, Unknown, "", ***, {}, n/a, N/A, NULL, null or not-set. Ensure all invalid values are removed from your index for this attribute.

Numerical type

code: numerical_type

This attribute is numerical across your index, making it less suitable as a categorical attribute. Ensure you use categorical attributes when configuring AI Personalization.

Mostly invalid values

code: mostly_invalid_values

Most of the values for this attribute in the index are invalid. Common examples of invalid values include unknown, Unknown, "", ***, {}, n/a, N/A, NULL, null or not-set. Ensure all invalid values are removed from your index for this attribute.

Excessive unique values

code: excessive_unique_values

Most values for this attribute in the index are unique, indicating it is not truly categorical. Ensure you use categorical attributes when configuring AI Personalization.

Single unique value

code: single_unique_value

All values for this attribute in the index are identical, but AI Personalization requires at least two categories. Ensure your index includes at least two distinct values for this attribute.

Did you find this page helpful?