Understanding personalization

The personalization feature in Sitecore Search customizes results by finding content items with similar values for attributes that you choose. You can use the Personalization feature in the Customer Engagement Console (CEC) to show results that are similar to a specific content item or to show results that are similar to content items a visitor has interacted with in the past. There are two types of personalization: More Like This (MLT) and Affinity.

MLT personalization uses specific documents and shows similar items in the results. MLT can either be based on document IDs that are specified in an API call or based on documents that the visitor has interacted with in the past. For example, Search can use the document ID of the current content item to show other similar content items. Alternatively, Search can use a visitor's UUID to find out what kind of documents they often interact with based on a specific attribute. If the visitor has interacted with blogs with the phrase cloud services in the title, MLT personalization shows them items that are like blogs about cloud services.

Affinity personalization does not use specific documents. Instead, it considers the visitor's browsing history and looks at metrics for different attributes. For example, this algorithm might learn that a visitor is interested in content items with the type blog based on their past views and clicks. The algorithm then boosts results that have type blog.

To set up the Personalization feature, first decide which attributes to use and enable them for use for personalization. Then, you configure options for the attribute in Feature Configuration and set up personalization options in the Global Widget or widget rules. Finally, you have to re-index your sources to ensure that the changes are implemented.

Note

If you are using MLT personalization with document IDs, a developer needs to send a personalization object to the API containing those IDs.

Configuring attributes for personalization

When you have chosen attributes for personalization and enabled them for personalization in the CEC, you need to configure those attributes on the Feature Configuration tab of the Domain Settings section of the Administration page of the CEC.

When you configure an attribute for use with personalization, the following options are available:

  • Analyzer - the analyzer to use on the attribute. In most cases, set this to Multi locale standard or lowercase.

  • Opt out of strategy - use this option to opt an attribute out of a personalization strategy. You might use this option to create two sets of personalization attributes for A/B testing.

Configuring personalization in the global widget or rules

When you configure Personalization in the Global Widget settings or in a rule, you first choose the personalization strategy. For both the MLT and Affinity strategies, the following option is available:

  • Attributes - the attributes that are used for personalization. Each attribute uses the analyzer configured in the Personalization section on the Feature Configuration tab of Domain Settings.

The following basic options are only available for the MLT strategy:

  • Last N Events - the number of past visitor events that are taken into consideration for personalization. The greater this number is, the longer it will take for personalization to react to visitors’ most recent actions. The recommended value range is 3-5. The default value is 5.

  • Boost Weight - the amount to boost personalized results when calculating results' relevancy scores. The greater this number is, the more personalization affects the final result set for a visitor. The recommended value range is 0.1-2.0. The default value is 0.5.

  • Minimum Number of Events - the minimum number of past visitor events required to start personalizing results. The greater this number is, the less reactive personalization will be. The recommended value range is 1-3. The default value is 2.

The following advanced options are only available for the MLT strategy:

  • Minimum Term Frequency - the minimum term frequency below which the terms will be ignored from the input document. The greater this number is, the less likely it is that personalization will find a pattern of common attribute value tags from your visitors’ PDP views. This causes personalization to act less frequently but with a higher relevance to visitors’ actual shopping intent. The default value is 1.

  • Minimum Document Frequency - the minimum document frequency below which the terms will be ignored from the input document. The greater this number is, the less likely it is that personalization will find a pattern of common attribute value tags from your visitors’ PDP views. This causes personalization to act less frequently but with a higher relevance to visitors’ actual shopping intent. The default value is 1.

  • Maximum Query Terms - the maximum number of query terms that will be selected. Increasing this value gives a broader personalization result set at the expense of query execution speed. The default is 8.

Do you have some feedback for us?

If you have suggestions for improving this article,