Here are some of the terms that are handy to know to better understand this documentation.
Revenue is counted as Anonymous Revenue if the user has opted out of being tracked during their visit, eg. by rejecting cookies in the browser. As a result, Anonymous Revenue will likely be a mix of Direct, Assisted, and Organic Revenue.
Revenue is counted as Assisted Revenue when the user uses recommendations to navigate the site before ultimately buying something different than what was recommended. If an order is placed in the same session as the user navigated to the page of a recommended product, and either navigates away from the page or adds the product to the basket before removing it again, an eventual order will count as Assisted Revenue.
Sometimes all users may not have access to all products on an e-commerce solution. That is often the case in a B2B scenario, and it is important that all personalization features adhere to this, so that, i.e., a product is not recommended to a user, that does not have access to it.
This functionality can be implemented using assortments. Assortments are a way to categorize products and users can be given access to assortments. All personalized elements will take into consideration if a user has access to an assortment containing a given product, and products a user does not have access to will never be recommended.
A category is a self-living entity that describes the relative position of another entity (product or content page) within the larger hierarchical tree structure of your website. This can be used for faceting, filtering, as well as for creating breadcrumbs on the PDP.
This is the sales channel where the recommendation, tracking, or search has been used. This could i.e. be your eCommerce webshop, an App, or an e-mail.
Classifications are metadata about the user. This could be the country the user is from or the customer segment. Relewise uses this information to make the recommendations and searches more relevant to the specific users as shopping behaviors are different in those Classifications.
All Currencies are supported in Relewise and there is no limit to the maximum number of currencies. A multicurrency data field consists of a key/value collection. The key is a string representing the currency and the value is a decimal representing the amount. There is no required format for a supplying string representing the currency.
A customer is someone who has purchased a license to use the Relewise recommendation engine. A customer is distinct from a User.
A dataset is where all of a customer's data is stored. This can include information about products, and content as well as tracking of behavioral information. You often have 2 datasets: One for production and one for testing. If you have multiple sites you might need more; please contact Relewise for advice on specific customer scenarios.
Decompounding is the act of splitting compound words into smaller searchable terms, eg. "doghouse" becomes "dog" and "house".
Debouncing refers to a time interval, usually 200-300ms, that is placed between input into a search field and a request for a predictive search result is sent to the Relewise API.
Revenue is counted as Direct Revenue when a user purchases one or more products that were directly recommended to them. This includes the user adding the recommended products to the basket as it is shown to them, as well as the user navigating to a recommended product, and then adding it to the basket. It is important to note that unless all recommended items are removed from the basket, the ensuing order will still count as Direct Revenue, even if the order is completed several days later.
Display Locations provide information to Relewise about where on your website your recommendations or searches have been shown to a user. This helps gather information about what parts of the website performs the best.
A collected term for Products/Product Categories and Content/Content Categories
Faceting is the act of narrowing search results on the client side. It is distinct from filters in that filtering happens in the call before the search or recommendation results are returned to the user, while facets are explicit options for the user to filter by.
Feeds allow you to import products and/or content pages (entitites) from your website or e-commerce platform into Relewise via a standardized format. Relewise innately supports RSS 2.0/Google Shopping Feeds, but can be made to work with any feed format you might have. Please note that importing a customized feed requires billed work, since we have to build the integration to fit your feed.
You can read more about Feeds here.
Filters are used to control what products or content are returned in recommendations or searches. This includes filtering by Assortments, product data fields, product IDs, and much more.
Filters for recommendations may be further customized through filter scopes, which allow you to specify what part of the recommendation results you want to apply the filter to.
Identifiers are keys that are used to identify a user in other systems. I.e. the key used in an e-mail marketing system or CRM system.
You can read more about identifiers here.
All languages are supported in Relewise and there is no limit to the maximum number of languages in a solution. This also means that languages that read left-to-right (LTR), as well as languages that read right-to-left (RTL) are supported.
Language information for language-specific strings is identified by a string key, typical examples would be "en", "dk" or "no". There is no required format for supplying these language codes (i.e. "en" vs. "en-us"). The important thing is that it is done in a consistent way.
Merchandising refers to a series of rules that can be applied to the Search and Recommendation algorithms, to help either boost, reduce or filter the occurrence of a product, a product category, a content page or similar.
Modifiers can be used to tweak the relevance calculations, use cases include but are not limited to:
- Increase the relevance of products with excess stock.
- Increase the relevance of seasonal products.
- Decrease the relevance of products that are nearly sold out and the chance that size matches the user is low.
- Increase the relevance of new products
Revenue is counted as Organic Revenue when the user did not use or engage with recommendations during their visit when the order was created.
The Power Step is a page or an overlay displayed when a user adds a product to the basket. On the Power Step are personalized product recommendations, showing cross-sell products for the product just added to the basket. It could also be products that are relevant to the users based on browsing and purchase history, top-selling products, or products on sale.
The Power Step is very effective, e.g. in the clothing industry, Relewise has seen customers have a conversion rate of approx. 50% on the Power Step.
Example of usage of Power Step
Product Details Page
The Products Details Page, or PDP, is the page that shows details such as description, pictures, specifications and price for a given product. Relewise uses the PDP to track user engagement and determine what products are trending and interesting in real time.
Relewise operates with several different distinct revenue types, namely Direct Revenue, Asisted Revenue, Organic Revenue and Anonymous Revenue. All Revenue tracking recorded in My Relewise is currently exclusively based on order-level, which is to say that if an order contains a product that would count as Direct Revenue, then the entire order is marked as such. Relewise does track products on line-level, and it will be possible to sort revenue by individual products in an order in a future release.
For Variants, specifications refer to the particular qualities that makes up each unique variant. A specification is thus something like
style. Specifications themselves are typically used for faceting.
Stemming refers to the common practice of rendering search words down to their base version, eg. "flying" becomes "fly". This makes it easier for the search engine to parse natural language and provide a broader scope of results from imprecise searches.
A set of data that tracks how an end-user of a website, uses the page (what products are visited/added to the basket, etc.)
A website end-user that is tracked in the system. Relewise distinguishes between three different types of users: Anonymous users, Temporary users, and Authenticated users.
Variants, or Variant Products, are products that specify certain qualities of a product, such as size, color, style, etc. Variant products are tied to an "umbrella" product, and have Specifications that identify their properties. Otherwise, they are treated identically to regular products for search and recommendation purposes.