Recommendation Strategies

This article lists all the different recommendation strategies available through the Hello Retail dashboard, and the logic behind each.  As of February 2025, these are all the individual algorithm steps, that you can select through the "Recommendations" menu:

Alternatives

This step selects similar products in relation to the currently viewed product.

The recommended products will be ordered by how similar they are to the currently viewed product. However, you can also choose to prioritize products whose price is close to the price of the currently viewed product. Similarity will still play the largest part, but for two potential recommended products that both have roughly the same similarity to the source product, you might prefer to show the one that is also similar in price. A slider is available, allowing you to define how much weight you want to put on this price similarity.

Bought together

Picks products which are often bought together with the currently viewed products.

By default, recommendations are ordered only by how often the products are bought together. As with "alternatives" above, you have the option of weighing similar prices as well.

Manual product selection

Picking the exact product you want to show. Products are chosen by searching for the title or URL in the search field.

Related products

Related products are the items, that the sum of all the visitors on the site are viewing most, together with the currently viewed product. 

Retargeted products

This strategy is based on the history of the individual user’s (cookie's) views. The first product in the product list is the most recently viewed product by the user, the next one is the second last viewed and so on.

Top products

These are the most popular products ranked according to how often a product is sold compared to how often it is viewed, by the sum of all visitors on the site.

Search

Normally, this is a search for some attribute in the title of a product, which is often the name of the product viewed by the customer. This step is not so much an algorithm, but more of an experimental approach to finding products that share some attributes with the search word.

Recently bought

Similar to the Retargeted products strategy, but focuses on the customer’s recently purchased products and not viewed products.

Current cart contents

This step returns the products that are present in the users' (cookie's) current cart. The cart is only considered current, if it has been updated within the last 20 minutes.

Most bought

Finds the products that are overall bought most often, by all the customers on the webshop. This measurement is further made into a dynamic value, by reducing the value over time, so that it represents what people are buying now, rather than only in the past. If many customers on the webshop are buying an item that can be considered seasonal, like for instance a Christmas sweater, then this item will over time become less likely to appear in the Most bought algorithm, unless people continue to buy it throughout the year of course. 

Most viewed

The products that the sum of all customers is visiting the most.

Products viewed together

This provides products viewed together (by the sum of all visitors on the shop) with either of two options that can be selected from the "Viewed with" dropdown menu: the user's recently viewed products or the user's recently bought products. Thereby it will either show products that many people viewed together with your own recently viewed products or products that many viewed together with what you recently bought.

Products bought together

This does the same as the algorithm above and provides the same 2 configuration options. The only difference is that the products returned are based on what the sum of all the other visitors are buying from the shop, and not what they are viewing.

Random products

Allows you to output a selected number of randomized products. This can be used for scenarios where the products need to be very dynamically changed, but not necessarily be related to other products.

Recently created

The output of this source is products sorted by creation date in descending order. It is possible to specify the creation date for each product by configuring this in the product feed, however, if that date is unavailable for a certain product, the default date is set to when the product was seen for the first time in the Hello Retail system.