Search and relevance in Avensia Storefront

Search and Relevance

The search engine is a very central component in Avensia Storefront Start Site. The search engine is used as a high-performance index by the software and by the visitors.

• The search engine is invoked when a visitor enters a search phrase to give auto complete suggestions to products that are available.
• If the visitor doesn’t select a suggested product but instead starts a search, is the search engine invoked to search for matching products.
• The search engine is also used to search web content, but that is not implemented by default in Avensia Storefront Start Site.
• Avensia Storefront Start Site renders a page with product information with products fetched from the search filter in a query sent to the search engine. The category page is a good example where products are filtered to be rendered on a web page.
• The search result is displayed using a category page in Avensia Storefront Start Site.

A standard search engine returns the result sets in a predefined sort order. In a solution using a merchandize software for relevance is the search result returned in a sort order based on relevance. That means that the most popular product for the current visitor is displayed top most in the result set. The algorithm behind the relevance and sort order is implemented by the merchandize service.

The merchandize service uses the behaviour of the visitor together with statistics for the visitor. It merges this information with behaviour and statistics of all visitors and gets a picture of what is most relevant. The merchandizer also ranks the components in the web pages to determine which component that is most used before an order is created. A sophisticated algorithm decides what components that should be placed on the web pages, and where on the web pages it should be placed.

The relevance engine uses virtual panels that are filled with relevant information. The panels are virtually placed on a web page in an order where the most ranked panel is placed top most. Panels are also used for recommendation such as “Those who bought this also bought”. The recommendation is based on behaviour and statistics.

Another recommendation is “The most popular products”. That recommendation is not based on the most sold product, but on the product that most visitors are interested in. The most sold product is probably some consumables.

The search engine in Avensia Storefront may be any search engine supported by Episerver. The search engine is indexed in the Episerver CMS Administration using one of two jobs. The default implementation of the jobs in Avensia Storefront is that they read product information from Episerver Commerce Catalog.

• Full search index. The search index is completely created from scratch. This option may be very time consuming and is only needed when the search index schema has been changed during development tasks.
• Incremental search index. Only changes in the Commerce Catalog are indexed. Products that are new, changed or removed are updated in the index. This is the fastest option to update the index and is the recommended option to use.

NOTE!

Since Avensia Storefront Start Site is Search Content Driven is it very important that the search index is up to date. A product will not be exposed in the web site until it is included in the search index.

Avensia Storefront uses the search engine and merchandizer to render web pages with product information and recommendations. The products selected by the merchandizer are sorted in a relevant order based on the visitor's behaviour in the web site and on all visitors behaviour in the web site. The web site sends behaviour to the merchandizer to keep the statistics up to date.