Introduced in Feature Pack 2

Search relevancy and merchandising

Search relevancy and merchandising is the process of controlling the search results that are returned to shoppers in the storefront, and the order in which they appear. There are several techniques that can be used to influence search relevancy, which lets you return products in the order that best suits your business needs.

Before you begin

Before working with search relevancy and merchandising to influence search results, ensure that your site meets the following requirements:
  1. Ensure that your Product Manager and merchandisers enter detailed product information into your store catalog. Well-chosen and clear product descriptions increase each product's visibility and relevancy in search results.
  2. Ensure that your catalog is indexed in WebSphere Commerce search, using any appropriate index types and subtypes. That is, your catalog data in the Catalog Entry index and your categories in the Catalog Group index. Optionally, and where applicable, your inventory data in the Inventory index subtype, and your price data in the Price index subtype.
  3. Ensure that your search implementation uses appropriate Solr and WebSphere Commerce-side indexing and querying features. That is, using appropriate index features such as field types, tokenizers, and filters; and using the appropriate query features such as search profiles, filters, pre-processors and post-processors.

After you have described your catalog and built the search indexes, and if you are still unsatisfied with search results in the storefront, you can fine-tune search relevancy to achieve your desired results.

Relevancy for keyword search

Relevancy for keyword search applies as a shopper searches the store.

Problems getting products returned

If you are having issues with the products being returned in search results, consider the following approaches:

If the products exist, but are not being returned:
  • Review the prerequisites in this topic to ensure that the product descriptions in your catalog are well-chosen and clear, that your content is indexed correctly, and that your search implementation uses all appropriate Solr and WebSphere Commerce-side indexing and querying features.
  • Create search term associations that use search terms that describe the products in your catalog to increase the relevancy of search terms.
  • Configure a second search to run, if the first search results in too few results. For example, by applying a different search type, or a different minimum match value to a second search query.
If too many search results are being returned in the storefront, which include both desired and irrelevant products:
  • Change the default search type to a more restrictive value to reduce the number of search results.
  • Use minimum match and phrase slop. Minimum match specifies the number of search keywords that are required to match the indexed document, when the ANY search type is used. Phrase slop specifies how far apart the indexed search terms are in the document to influence relevancy.
  • Use category-based search relevancy to filter products that are displayed to shoppers on the search results page.
  • Limit auto suggestions and spell correction to a specific catalog.
  • Limit specific search terms and characters from the search query such as unimportant words, stemming, or disabling wildcards and other characters. That is, work with stop words, protected words, and prohibited words to limit the amount of searchable words in the storefront.
If too few search results are being returned in the storefront, consider the following approaches:
  • Change the default search type to a less restrictive value to increase the number of search results.
  • Create search term associations that use search terms that describe products in your catalog to increase the scope of search terms.
  • Enable and work with auto suggestions to ensure that shoppers find products that match partial search terms.
  • Work with fuzzy search so that additional search results are displayed that might relate to the search terms.
If you want more control over which products are being returned in the storefront, and their ordering, consider the following approaches:
  • Use landing pages and search rules such as Specify Top Search Result to deliver customized search results and ordering.
  • Work with search result grouping to generate search results and achieve visual relevancy by showing the product's relevant attributes (such as color) first.

Problems with product sequencing

If you want more control over which products are being returned in the storefront, and their ordering, consider the following approaches:
  • Use Solr-specific functions to control the sequence of returned products, such as the ngram, omitNorms, and omitTermFreqAndPosition parameters.
  • Change the relevancy of search index fields to match how your catalog is described. Review the default field boosting and change or boost the fields that contain the most relevant information.
  • Use search rules to promote and demote products that are returned in search results.
  • Boost the shopper's search query higher than search term associations.
  • Use category-based search relevancy to boost products that are displayed to shoppers on the search results page.

Relevancy for site navigation (search merchandising)

Relevancy for site navigation applies as a shopper navigates the store. Shoppers are presented with products that might have been boosted, promoted, sorted or merchandised in the storefront.
  • Use category-based search relevancy to filter products to show results from specific categories.
  • Control product sequencing by using deep search sequencing and deep category unpublishing.
  • Use Exclusive and Clearance search facts to feature products.
  • Create rule-based sales categories in a sales catalog to quickly and efficiently promote seasonal sales categories.
  • Review site search statistics to make targeted improvements to optimize your site search, improve your site navigation, and ultimately increase conversions.
  • Use inventory data to promote overstock products and hide out-of-stock products.
  • Use product data to dynamically recommend products to customers based on profit margin.