Relewise Documentation

Appearance

Implementation and Go-Live Checklist ​

General ​

  • Ensure that only production datasets are used in production.
  • All ids are case sensitive so when checking ids for entities, ensure the casing also matches in Relewise.
  • If requests to Relewise are being made directly in the browser, ensure that the api key does not have write or other dangerous permissions.
  • Ensure all users see a cookie consent banner the first time they hit the site, regardless of what page they first hit.

Behavioral Tracking ​

  • User ​

    • If cookie consent (for marketing) is not given, ensure the user is anonymous.
    • When using segmentation always include classifications as they are not personal identifiable and do not require consent.
    • Make sure you have consultet Relewise before applying different segmentations. Most use cases only need "Country" as a segmentation. Too many classifications can water down the personalization leading to poor results.
    • Include all identifiers known for the user.

  • Product view ​

    • The user viewing the product is included.
    • Whenever a product page is rendered, a product view is tracked.
    • The product id is included and it matches a product in Relewise.
    • When viewing a variant, ensure that the variant's id is also included and matches a variant on the product in Relewise.
    • When changing variant make sure to track a new product view
    • Product views should only be used on Product Details Pages, not when viewing products other places

  • Product category view ​

    Only relevant if products have categories.

    • The user viewing the category is included.
    • Whenever a product category page is rendered, a product category view is tracked.
    • The product category id is included and it matches a category entity in Relewise.

  • Content view ​

    Only relevant if the site contains content.

    • The user viewing the content is included.
    • Whenever a content page is rendered a content view is tracked.
    • The content id is included and it matches a content entity in Relewise.

  • Content category view ​

    Only relevant if the site contains content and the content contains categories.

    • The user viewing the category is included.
    • Whenever a content category page is rendered a content category view is tracked.
    • The content category id is included and it matches a category entity in Relewise.

  • Brand view ​

    Only relevant if the site contains multiple brands.

    • The user viewing the brand is included.
    • Whenever a brand page is rendered a brand view is tracked.
    • The brand id is included and it matches a brand in Relewise.

  • Cart update ​

    • The user the cart is connected to.
    • When a product or variant is added to the cart, an update cart request is sent.
    • When a product or variant is removed from the cart, an update cart request is sent.
    • Changes to the quantity of one specific line item in the cart should also trigger an update.
    • Make sure the basket contains all line items in every requests.
    • If it is necessary to be able to identify a basket (fx. for triggers) ensure some recognizable id is stored in basket data.

  • Search Term ​

    Only relevant when not using Relewise search as this then is tracked implicitly.

    • The user the performing the search.
    • The language in which the search is performed should always be included.
    • What term the user actually has searched for should also be provided.

Recommendations ​

  • The user viewing the recommendation should be included.

  • The number of entities returned by Relewise should all be rendered on the page.

  • The language in which the recommendation is displayed should be included in the request.

  • When displaying products the currency should be included as well.

  • Where the recommendations are being rendered should be included in the Displayed At Location

  • When rendering multiple recommendations on the same page, these recommendations should be batched

  • Make sure that the order of results are maintained. The recommendations are returned in order of most relevant first. If using a different system to fetch product data, make sure that the order of the recommendations are not changed when fetching the data from a different system

  • Purchased with Product ​

    • The id of the product to base the recommendation on should always be included.
    • If the recommendation is based on a variant the variant's id should also be included.

  • Purchased with multiple Products ​

    • The ids of the products on which the recommendation should be based should always be included.
    • If the recommendation is based on a variant, the variant ID should also be included.

  • Purchased with current Cart ​

    • This recommendation should generally not be used as it only works for known users, so to simply the implementation consider only implementing Purchased with multiple Products
    • It is especially important that cart updates are being tracked when using this recommendation.
    • The type of popularity should be included. It should either be Most Purchased or Most Viewed.
    • How many minutes back in time the recommendation can use data from to base the popularity on.
    • The wieghts should be provided for Category views, Product views, and Product purchases. The number value represents the factor of how important the datapoint should be for the popularity.
    • How many minutes back in time the recommendation can use data from to base the popularity on.
    • When displaying this recommandation in the context of a user search, the term the user is searching for should be included.

  • Popular Content ​

    • How many minutes back in time the recommendation can use data from to base the popularity on.

  • Popular Content categories ​

    • How many minutes back in time the recommendation can use data from to base the popularity on.
    • The wieghts should be provided for Category and Content views. The number value represents the factor of how important the datapoint should be for the popularity.
    • How many minutes back in time the recommendation can use data from to base the popularity on.
    • The wieghts should be provided for Brand views, Product views, and Product purchases. The number value represents the factor of how important the datapoint should be for the popularity.

  • Sort Products ​

    • A list of product ids Relewise should sort, should be included. Ensure all product IDs exist for products stored in Relewise.

  • Sort Variants ​

    • The ID of the product and which variants should be sorted. Ensure a product exists and that it has variants.

  • Personal Product ​

    • Including the same user that is used for tracking is always important, but especially here. Please double check it is correct.

  • Personal Product Category ​

    • Including the same user that is used for tracking is always important but especially here. Please double check it is correct.
    • The wieghts should be provided for Category views, Product views, and Product purchases. The number value represents the factor of how important the datapoint should be for the popularity.

  • Personal Content ​

    • Including the same user that is used for tracking is always important, but especially here. Please double check it is correct.

  • Personal Content Category ​

    • Including the same user that is used for tracking is always important but especially here. Please double check it is correct.
    • The wieghts should be provided for Category and Content views. The number value represents the factor of how important the datapoint should be for the popularity.

  • Personal Brand ​

    • Including the same user that is used for tracking is always important but especially here. Please double check it is correct.
    • The wieghts should be provided for Brand views, Product views, and Product purchases. The number value represents the factor of how important the datapoint should be for the popularity.

  • Products viewed after viewing Product ​

    • The id of the product to base the recommendation on should always be included.
    • If the recommendation is based on a variant the variant's id should also be included.

  • Products viewed after viewing Content ​

    • The id of the content to base the recommendation on should always be included.

  • Content viewed after viewing Product ​

    • The id of the product to base the recommendation on should always be included.
    • If the recommendation is based on a variant the variant's id should also be included.

  • Content viewed after viewing multiple Products ​

    • The ids of the products to base the recommendation on should always be included.
    • If the recommendation is based on a variants the variants ids should also be included.

  • Content viewed after viewing Content ​

    • The id of the content to base the recommendation on should always be included.

  • Content viewed after viewing multiple Content ​

    • The ids of the content to base the recommendation on should always be included.

  • Recently Viewed Products ​

    • Including the same user that is used for tracking is always important but especially here. Please double check it is correct.

  • Search Term-Based Product ​

    • The term the user is searching for should always be provided.

  • Similar Products ​

    • The id of the product to base the recommendation on should always be included.
    • If the recommendation is based on a variant the variant's id should also be included.
    • The significance of datakeys used for calculation similarity should be included.

Search Types

The following checklist applies to all types of search that you are likely to engage in. As such, you should run through the checklist once for each of the following search types, as applicable:

  • Ensure a search index is created and correctly configured.
  • Make sure that the order of results on the result page is shown in the same order as we returned them. The results are returned in order of most relevant first. If using a different system to fetch product data, make sure that the order of the results are not changed when fetching the data from a different system.
  • The user performing the search should be included.
  • The number of entities returned by Relewise should be rendered on the page.
  • The language in which the search is performed should be included in the request.
  • When displaying products the currency should be included as well.
  • Where the search is being performed should be included in the Displayed At Location
  • When performing multiple searches on the same page, these searches should be batched. This could be a Product Search Request along with a Search Term Prediction Request.
  • The number of entities the result should include should be provided and should stay consistent when pagination.
  • The number of entities to skip should match the number of products displayed times the page number.
  • When the user has the option to include a term, that term should always be included.
  • When searching for a term a debounce time should be implemented so every key press does not trigger a request.
  • When applying a sort order, make sure to always perform a then by Product Relevance afterwards to make sure Relewise personalizes the different buckets in the sorting

Don't know us? Don't worry - you can find more information about us, by visiting our main page www.relewise.com

ContactChangelogsCookies, Privacy & TermsAPI Reference