A product recommendation strategy comprises of a primary strategy and an optional fallback strategy.

Primary strategy is the main strategy that is used to generate product recommendations for the selected zone. In addition to the primary strategy, you can also create a fallback strategy that comes into play if the primary strategy doesn’t generate sufficient number of relevant product recommendations. When creating a fallback strategy, you cannot choose exactly the same set of parameters as in the primary strategy. However, you can use the same business rules as the primary strategy or you can also select new business rules.

When a product recommendation strategy is published, both primary & fallback strategies become active. The output of a product recommendation strategy is a combination of the results from both, the primary and fallback strategy.  

Create a primary strategy

To create a primary recommendation strategy, complete the following steps:

1. Go to Zones and select the channel for which you want to configure product recommendations.
2. In the zone for which you want to create recommendations, click Actions > Create recommendation.
   
3. Click Create recommendation strategy. 
4. Provide a name for the strategy.
5. The Product Catalog drop-down shown all the product catalogs available for the organization.  Select the product catalog relevant to your channel from the list. 
   
6. Click Create strategy. The Primary strategy tab opens by default, showing the strategy name and the catalog name. 
   
7. In the Recommendations title field, you can optionally specify a custom heading (title) for the recommendation zone on your channel. This enables you to communicate the objective of each recommendation strategy to your website visitors, and be more transparent about what you are recommending to your users and why.
   Here are some suggested header text options to get you started:
   To use this field, ensure that Personalization library version 2.2.0 (or later) is deployed on your site and the zones are configured. For more details, check out this article in the Developer Center.
   This field is optional. You can leave it empty or update titles for your recommendations manually on your website if you prefer.
8. In the Model section, select the algorithm for the product recommendations. For more details about the algorithms, see Algorithms 
   If you select Most popular algorithm, then you also need to select the measure of popularity parameter. This generates recommendations across all the categories available in the given time frame of the behavioral data. 
   
9. In the Lookback time frame field, select the number of days of historical visitor behavior data that should be used by the algorithm to generate product recommendations. You can set the value for Lookback time frame to either 1 day, 7 days, or for 30 days. 
   For a high-traffic website, the lookback period of 1-day would give good recommendations. Otherwise, you should select 7 days or 30 days, depending on the number of the unique visitors to your website. When choosing this value, you need to consider the trade-off between quality of data and the required quantity of data. Recent data is usually of more relevance and hence you may want to choose a value of 1-day or 7-days. However, it is possible that you may not have enough data for the models, and hence may need to choose a longer time frame.
10. If you had selected the Most popular as the algorithm, select the Measure of popularity, such as the most popular products based on view count, buy count, margin, and revenue. 

Specify model inputs

You need Personalization library version 2.1.0 or later configured to use this feature.

In the Type of input data field, select the input type that you want to use as the basis for product recommendations.

Notes 

• If you select algorithm as View to Buy OR Abandon to Buy, then the Model input field only shows  Product ID as option. This is because these two algorithms require product IDs as inputs. 
• If you select algorithm as Most popular, the following options are not available:
  • For the Type of input data as Web page, the Model input field does not show Product ID as an option.
  • For the Type of input data as User behavior, the Model input field does not have the following options: 
    • Most recently viewed products(s)
    • Most recently purchased products(s)
    • Most recently carted products(s)
    • Largest purchase
• If you select Type of input data as Web page and Model input as Attributes, then in the Input attributes field, select the attributes in the order they should be processed. 

For tips and guidance about how you can use these model inputs on various pages of your channel for optimal recommendations experience, see the section Model placement.

Business rules

You can specify business rules to define conditions that a product must fulfill to be part of the final recommendations. To add business rules, perform the following steps:

1. Click Add rule.
2. Provide a name to the rule. The business rule name must be unique within the strategy.
3. Choose one of the following options:
   • Apply this business rule to all the products in the catalog
   • Apply the business rule to specific products: In this case, specify the criteria using the condition builder tool.
     
     
     Select an Attribute. Select the Operator. Enter the value for attributes. The attributes are same as ones uploaded to Personalization using the Product catalog. If you had selected the algorithm as Most popular, then this option is not available.
4. Applying business rule to a category within a hierarchy

   Consider the following scenario: You have a product catalog with the following categories:

   Clothing
   	Menswear
   		Shirts
   			Formals
   				Suites
   			Casuals
   				T-shirts
   		Trousers

   You can create a business rule by specifying the precise hierarchy of the product category that you want to use for applying the business rule. For example, you can create a business rule that says, “If the Category is Clothing.Menswear.Shirts.Formals then Boost products with Category as Tuxedos.”

   To use the category hierarchy, perform the following steps:
   From the attributes list, select Category. Select the Value field as Specific Value. 
   
   
   The Select categories field shows a tree-view hierarchy of product categories, based on your product catalog.
   
   Choose one or more categories for which you want to apply the business rule.  

5. In the Outcome field, specify the action to be performed on the products that match the specified criteria. For more details about the actions (Pin, Bury, Boost, Filter, and Exclude), see Business rules.

   

    
6. Click Save to add the business rule. 
   To edit or delete a business rule of a published strategy, you need to first unpublish the strategy.
   • To edit business rule: Click Edit condition. You can update the Attribute, Operator, and Value fields, and then click Save.
   • To delete a business rule: You can delete the business rules that are added in your product recommendation strategy. To delete a business rule: Click Delete condition. Deleted rules are removed permanently.
7. In the product recommendations strategy page, click Save. Saving the strategy does not make it active on the channel.

You can preview the strategy before publishing it. If you want to publish the strategy on the channel, without previewing it, click Publish.

Create fallback strategy

Fallback strategy is used if the primary strategy does not generate any product recommendations for a product or if the primary strategy does not return sufficient recommendations for the zone. For example, if a website needs 10 product recommendations, but the primary strategy provides only 5, the fallback strategy will provide the remaining five recommendations.

When creating a fallback strategy, you cannot choose exactly the same set of parameters as in the primary strategy. However, you can use the same business rules as the primary strategy, or you can create new business rules.

Creating a fallback strategy ensures your visitors always have relevant and sufficient number of product recommendations on your site.

To create a fallback strategy, complete the following steps:

1. Click Fallback strategy tab and provide the required details.
2. From the algorithm list, select either None or Most popular. The Most popular algorithm is used to show the most popular products based on measure of popularity such as view count, purchase count, and revenue.
3. In the Lookback time frame field, select the number of days of historical visitor behavior data that should be used by the algorithm to generate product recommendations. You can set the value for Lookback time frame to either 1 day, 7 days, or for 30 days. 
   
   
4. Select the Measure of popularity, such as the most popular products based on view count, purchase count, margin, and revenue. 
5. Create business rules in one of the following ways:
   • Copy business rules from the primary strategy by clicking Use business rules from primary strategy check box. This saves you the hassle of creating the same business rules again for the Fallback strategy.
   • Click Add rule to create a new business rule.
6. Click Save.

Saving the strategy does not make it active on the channel. You can preview the strategy before publishing it. If you want to publish the strategy without previewing it, click Publish.