How Retail Rocket Group’s recommendation algorithms help people find what they need
How Retail Rocket Group’s recommendation algorithms help people find what they need
We call product recommendations a set of widgets with a selection of products, which are placed on a website, in an application or in an e-mail. They are designed to help users find what they need as quickly as possible among a huge range of products
34 algorithms that will replace a live consultant salesperson
- Match the product with alternative accessories
- Help the customer to find the right product faster
- Emphasise sales hits, promotional items and new products
- Increase the cheque by offering a product with improved features
Recommendations can be used in almost any sales channel: on the website, in the app, in email newsletters, on offline checkout screens and on sales assistants tablets. They are designed to give your users a more enjoyable shopping experience, while you grow LTV and your profits.
Popular Products
6 algorithms
Basic algorithm
Basic algorithm
Available solutions:
Recommendations of the most popular products are formed on the basis of all user actions in the online shop. The algorithm shows the products that are most often viewed/added to favourites and basket, and most importantly, bought.
The popularity of categories also influences the formation of recommendations. The algorithm shows the most diverse products that are favoured by other users. This helps to get better acquainted with the shop`s product base, as well as facilitates the selection process.
Where we recommend using the algorithm
Pages
Home page
404 page
User account
Empty search
Letters
Abandoned browsing in the popular products category
Popular in category
Popular in category
Available solutions:
A variant of the "Popular Products" algorithm, where the user is shown sales hits from a certain category. If it contains subcategories (for example, for the category "outerwear" it could be "coats", "fur coats"), the algorithm will show as many different subcategories as possible. Among them, the visitor is sure to find something that interests him.
Thanks to this scenario, the user doesn`t have to browse through several dozens of items from the category he needs - he can save time by choosing from the most popular products.
A similar algorithm is available via API, which returns an increased number of products (up to 2000) sorted by decreasing popularity.
Where we recommend using the algorithm
Pages
Home page
404 page
User account
Empty search
Hits of a particular brand
Hits of a particular brand
Available solutions:
The algorithm is used to promote currently popular products of a certain brand, sorted by decreasing popularity. The algorithm can generate output both for all products of this brand and for products from the specified categories.
Where we recommend using the algorithm
Pages
Home page
Product card
404 page
User account
Letters
Brand page
Brand Category
Category
Popular products with discounts
Popular products with discounts
Available solutions:
The algorithm suggests the most popular products of the shop at the moment, for which there is a discount. It takes into account clicks, views, add to basket, orders and other events. This algorithm helps users to pay attention to favourable offers.
Where we recommend using the algorithm
Pages
Home page
404 page
Category
New
Letters
Abandoned category view
Popular products from categories of interest to the user
real-time & personal
Popular products from categories of interest to the user
real-time & personal
Available solutions:
The algorithm remembers in real time the categories and products that were of interest to a particular user.
Interest in a category depends on the length of the consumption cycle. For example, purchases of regular demand goods (food, pet food) increase the interest in the category, as there is a high probability that next time the user will make a similar purchase. Buying large household appliances, furniture and other goods with a long consumption cycle decreases interest in the category.
The algorithm gives out hits from the categories that are most interesting to a given user in the long term, taking into account their price preferences. The output depends on the stage of the funnel. For example, if a user bought a product from the "Mobile Phones" category, the algorithm will show him popular products from related categories - "Phone Cases", "Headsets", "Mobile Phone Accessories". And if there is no product browsing history or interest in categories, popular products without personalisation will be shown.
When using the algorithm in emails, recommendations are built by categories popular with a given user.
Where we recommend using the algorithm
Home page
404 page
Category
New
Empty search
Empty cart
Popular products with discounts from categories of interest to the user
real-time & personal
Popular products with discounts from categories of interest to the user
real-time & personal
Available solutions:
A variation of the algorithm "Popular items from categories of interest to the user", only discounted items are offered.
Where we recommend using the algorithm
Home page
Promotions/Sales
404 page
Basic algorithm
Available solutions:
Recommendations of the most popular products are formed on the basis of all user actions in the online shop. The algorithm shows the products that are most often viewed/added to favourites and basket, and most importantly, bought.
The popularity of categories also influences the formation of recommendations. The algorithm shows the most diverse products that are favoured by other users. This helps to get better acquainted with the shop`s product base, as well as facilitates the selection process.
Where we recommend using the algorithm
Pages
Home page
404 page
User account
Empty search
Letters
Abandoned browsing in the popular products category
Popular in category
Available solutions:
A variant of the "Popular Products" algorithm, where the user is shown sales hits from a certain category. If it contains subcategories (for example, for the category "outerwear" it could be "coats", "fur coats"), the algorithm will show as many different subcategories as possible. Among them, the visitor is sure to find something that interests him.
Thanks to this scenario, the user doesn`t have to browse through several dozens of items from the category he needs - he can save time by choosing from the most popular products.
A similar algorithm is available via API, which returns an increased number of products (up to 2000) sorted by decreasing popularity.
Where we recommend using the algorithm
Pages
Home page
404 page
User account
Empty search
Hits of a particular brand
Available solutions:
The algorithm is used to promote currently popular products of a certain brand, sorted by decreasing popularity. The algorithm can generate output both for all products of this brand and for products from the specified categories.
Where we recommend using the algorithm
Pages
Home page
Product card
404 page
User account
Letters
Brand page
Brand Category
Category
Popular products with discounts
Available solutions:
The algorithm suggests the most popular products of the shop at the moment, for which there is a discount. It takes into account clicks, views, add to basket, orders and other events. This algorithm helps users to pay attention to favourable offers.
Where we recommend using the algorithm
Pages
Home page
404 page
Category
New
Letters
Abandoned category view
Popular products from categories of interest to the user
real-time & personal
Available solutions:
The algorithm remembers in real time the categories and products that were of interest to a particular user.
Interest in a category depends on the length of the consumption cycle. For example, purchases of regular demand goods (food, pet food) increase the interest in the category, as there is a high probability that next time the user will make a similar purchase. Buying large household appliances, furniture and other goods with a long consumption cycle decreases interest in the category.
The algorithm gives out hits from the categories that are most interesting to a given user in the long term, taking into account their price preferences. The output depends on the stage of the funnel. For example, if a user bought a product from the "Mobile Phones" category, the algorithm will show him popular products from related categories - "Phone Cases", "Headsets", "Mobile Phone Accessories". And if there is no product browsing history or interest in categories, popular products without personalisation will be shown.
When using the algorithm in emails, recommendations are built by categories popular with a given user.
Where we recommend using the algorithm
Home page
404 page
Category
New
Empty search
Empty cart
Popular products with discounts from categories of interest to the user
real-time & personal
Available solutions:
A variation of the algorithm "Popular items from categories of interest to the user", only discounted items are offered.
Where we recommend using the algorithm
Home page
Promotions/Sales
404 page
New
4 algorithms
Basic algorithm
Basic algorithm
Available solutions:
The algorithm recommends products sorted by date of arrival, starting from the newest ones. It can generate output for all products of the shop, as well as for the specified categories.
The algorithm is useful for online shops in segments where the assortment is frequently replenished: fashion, online cinemas, digital libraries, music streaming, home and leisure products, tickets.
Where we recommend using the algorithm
Pages
Home page
Category
New
404 page
Letters
Abandoned category view
New products with discounts
New products with discounts
Available solutions:
The algorithm selects new products with discounts. The script helps the user not to miss out on a favourable offer, and helps the online shop to show how it differs from its competitors. The algorithm can generate output both for all the shop`s products and for the specified categories.
Where we recommend using the algorithm
Pages
Home page
User account
Empty search
404 page
Letters
Abandoned browsing in a category
In any email: popular across the shop
New products from interesting user categories
real-time & personal
New products from interesting user categories
Available solutions:
A personalised variant of the "Novelties" algorithm allows you to show recently available products only from the categories that are most interesting to a given user in the long term.
For example, if a user was looking at products from the categories "T-shirts" and "Shorts", the algorithm will suggest new products from these categories. And if he bought a product from the "Mobile Phones" category, the algorithm will offer recently arrived products from related categories: cases, headphones and other accessories.
New visitors will see novelties without personalisation.
Where we recommend using the algorithm
Home page
New
404 page
New products with discounts from categories of interest to the user
real-time & personal
New products with discounts from categories of interest to the user
Available solutions:
A variation of the personalised version of the "Novelties" algorithm, which shows the user the most interesting discounted offer from a new range of products.
Where we recommend using the algorithm
Home page
Promotions/Sales
404 page
New
Basic algorithm
Available solutions:
The algorithm recommends products sorted by date of arrival, starting from the newest ones. It can generate output for all products of the shop, as well as for the specified categories.
The algorithm is useful for online shops in segments where the assortment is frequently replenished: fashion, online cinemas, digital libraries, music streaming, home and leisure products, tickets.
Where we recommend using the algorithm
Pages
Home page
Category
New
404 page
Letters
Abandoned category view
New products with discounts
Available solutions:
The algorithm selects new products with discounts. The script helps the user not to miss out on a favourable offer, and helps the online shop to show how it differs from its competitors. The algorithm can generate output both for all the shop`s products and for the specified categories.
Where we recommend using the algorithm
Pages
Home page
User account
Empty search
404 page
Letters
Abandoned browsing in a category
In any email: popular across the shop
New products from interesting user categories
Available solutions:
A personalised variant of the "Novelties" algorithm allows you to show recently available products only from the categories that are most interesting to a given user in the long term.
For example, if a user was looking at products from the categories "T-shirts" and "Shorts", the algorithm will suggest new products from these categories. And if he bought a product from the "Mobile Phones" category, the algorithm will offer recently arrived products from related categories: cases, headphones and other accessories.
New visitors will see novelties without personalisation.
Where we recommend using the algorithm
Home page
New
404 page
New products with discounts from categories of interest to the user
Available solutions:
A variation of the personalised version of the "Novelties" algorithm, which shows the user the most interesting discounted offer from a new range of products.
Where we recommend using the algorithm
Home page
Promotions/Sales
404 page
New
Alternative products
6 algorithms
Basic algorithm
Basic algorithm
Available solutions:
The algorithm suggests to the user products similar to the one he/she is currently looking at. The selection is formed on the basis of product descriptions and properties, as well as on the basis of the behaviour of other users who are interested in the same product: what else they are studying and buying. Therefore, the algorithm can suggest a product that is not always similar in description, but really suitable.
Where we recommend using the algorithm
Pages
Product card
Comparison page
Letters
Abandoned browsing
Abandoned cart
Upsell
Upsell
Available solutions:
In offline shops, improved and more expensive alternatives are usually suggested by consultants. The Upsell algorithm replaces such a salesperson online. It identifies the implicit need of a person and helps the user to fulfil their request better and the online shop to earn more. This algorithm recommends products that are as similar as possible, but with improved features and more expensive.
For example, if a user is looking at wired headphones for 2,000 roubles, they will be shown very similar but wireless headphones for 3,000 roubles. The buyer can compare them, calculate that it will be more comfortable without a wire, and eventually buy the more expensive version.
Where we recommend using the algorithm
Product card
Cart
Margin-optimised alternative products
Margin-optimised alternative products
Available solutions:
A variant of the "Alternative Products" algorithm, in which the user is shown the most marginal of the products that are suitable for the user.
The algorithm can reduce a shop`s conversion rate while maintaining or increasing marginal revenue. This is partly by reducing the number of orders and their associated costs.
Where we recommend using the algorithm
Pages
Product card
Letters
Abandoned browsing
Abandoned cart
Revenue-optimised alternative products
Revenue-optimised alternative products
Available solutions:
A variant of the Alternative Products algorithm in which similar products are selected in such a way as to maximise the predicted revenue per user (RPV).
The algorithm can help to increase the average cheque.
Where we recommend using the algorithm
Product card
Analogues to missing items
Analogues to missing items
Available solutions:
The algorithm recommends to "out-of-stock" items that are as similar as possible in terms of product characteristics, but which are available for purchase.
The algorithm differs from alternatives in that it selects products with properties as similar as possible to the out-of-stock product, while alternatives show similar products, taking into account user behaviour.
Where we recommend using the algorithm
Card of unavailable product
Cart
Similar products with emphasis on the products viewed by the user
Similar products with emphasis on the products viewed by the user
Available solutions:
A personalised version of the "Alternative Products" algorithm, in which previously viewed products are displayed on the first places in the output. A block with this algorithm allows the user to remember what he or she has looked at before, compare it with the current product and decide on a choice.
Where we recommend using the algorithm
Product card
Basic algorithm
Available solutions:
The algorithm suggests to the user products similar to the one he/she is currently looking at. The selection is formed on the basis of product descriptions and properties, as well as on the basis of the behaviour of other users who are interested in the same product: what else they are studying and buying. Therefore, the algorithm can suggest a product that is not always similar in description, but really suitable.
Where we recommend using the algorithm
Pages
Product card
Comparison page
Letters
Abandoned browsing
Abandoned cart
Upsell
Available solutions:
In offline shops, improved and more expensive alternatives are usually suggested by consultants. The Upsell algorithm replaces such a salesperson online. It identifies the implicit need of a person and helps the user to fulfil their request better and the online shop to earn more. This algorithm recommends products that are as similar as possible, but with improved features and more expensive.
For example, if a user is looking at wired headphones for 2,000 roubles, they will be shown very similar but wireless headphones for 3,000 roubles. The buyer can compare them, calculate that it will be more comfortable without a wire, and eventually buy the more expensive version.
Where we recommend using the algorithm
Product card
Cart
Margin-optimised alternative products
Available solutions:
A variant of the "Alternative Products" algorithm, in which the user is shown the most marginal of the products that are suitable for the user.
The algorithm can reduce a shop`s conversion rate while maintaining or increasing marginal revenue. This is partly by reducing the number of orders and their associated costs.
Where we recommend using the algorithm
Pages
Product card
Letters
Abandoned browsing
Abandoned cart
Revenue-optimised alternative products
Available solutions:
A variant of the Alternative Products algorithm in which similar products are selected in such a way as to maximise the predicted revenue per user (RPV).
The algorithm can help to increase the average cheque.
Where we recommend using the algorithm
Product card
Analogues to missing items
Available solutions:
The algorithm recommends to "out-of-stock" items that are as similar as possible in terms of product characteristics, but which are available for purchase.
The algorithm differs from alternatives in that it selects products with properties as similar as possible to the out-of-stock product, while alternatives show similar products, taking into account user behaviour.
Where we recommend using the algorithm
Card of unavailable product
Cart
Similar products with emphasis on the products viewed by the user
Available solutions:
A personalised version of the "Alternative Products" algorithm, in which previously viewed products are displayed on the first places in the output. A block with this algorithm allows the user to remember what he or she has looked at before, compare it with the current product and decide on a choice.
Where we recommend using the algorithm
Product card
Related products
8 algorithms
Basic algorithm
Basic algorithm
Available solutions:
The algorithm recommends products that complement the user`s order. For example, when buying an inflatable pool, it can immediately recommend a pool cleaner.
When there is insufficient data on user behaviour, the algorithm shows products that can be purchased together based on their properties, category membership and popularity.
The algorithm helps the user to familiarise themselves with the shop`s range and not to forget necessary parts or ingredients.
Where we recommend using the algorithm
Pages
Product card
Cart
Letters
Post-transaction
Accessories
Accessories
Available solutions:
A variation of the "Related Products" algorithm. Matches additional accessories to the product being viewed. Helps the user to close his need in one place. For example, to pick up a case, headphones and protective glass for a smartphone at once, which increases the average cheque of the online shop.
Unlike the basic algorithm, priority is given to products from categories that we define ourselves based on our experience, as well as products of the same brand.
Where we recommend using the algorithm
Pages
Product card
Cart
Letters
Post-transaction
Related products prioritised by brand of the product being viewed
Related products prioritised by brand of the product being viewed
Available solutions:
A variant of the Related Products algorithm. It is used when there is not enough information about user behaviour.
For example, based on behavioural statistics, we see that a particular chair is bought together with a particular table, then the recommendations will first of all suggest it. But if we don`t know what is usually bought with an Eames dining table, then the algorithm will recommend chairs and other furniture of the same brand.
Where we recommend using the algorithm
Product card
Related products from categories, other than the current product category
Related products from categories, other than the current product category
Available solutions:
A variant of the "Related Products" algorithm, in which products of the same category as the product being viewed are removed from recommendations.
For example, it makes no sense for an electronics shop to recommend other TV sets to a TV set. But if we are talking about baby food, then apple puree in addition to pear puree would be organic.
Where we recommend using the algorithm
Product card
Cart
Accessories tailored to the shop`s requirements
Accessories tailored to the shop`s requirements
Available solutions:
Complements the basic "Related Products" algorithm and allows the shop to adjust product rendition using manually configurable rules. It will be useful if you need to promote products from unpopular categories or the shop wants to ensure that only really suitable related products are recommended to the main product. Relevant for technically complex products: snowmobiles, computers and so on.
Where we recommend using the algorithm
Product card
Cart
Related categories
Accessories tailored to the shop`s requirements
Available solutions:
The algorithm displays up to 20 related categories for a given category, sorted in descending order of popularity. For example, for washing machines, this could be powders, descalers and accessories.
The algorithm is used to simplify navigation: the customer is presented with categories where he can find additional products for his purchase.
Accessories with high diversity requirements
Accessories tailored to the shop`s requirements
Available solutions:
A variation of the "Accessories" algorithm with a more diverse output. It is suitable if there are a lot of accessories that are very similar in terms of product properties.
Where we recommend using the algorithm
Product card
Cart
Related products to the last added to basket
Related products to the last added to basket
Available solutions:
The algorithm is designed for use on product pages. It analyses the contents of the shopping cart and suggests products that can complement the purchase. An attempt to sell related products before the user proceeds to checkout can be more effective than a similar one on the "Basket" page.
Where we recommend using the algorithm
Product card
Basic algorithm
Available solutions:
The algorithm recommends products that complement the user`s order. For example, when buying an inflatable pool, it can immediately recommend a pool cleaner.
When there is insufficient data on user behaviour, the algorithm shows products that can be purchased together based on their properties, category membership and popularity.
The algorithm helps the user to familiarise themselves with the shop`s range and not to forget necessary parts or ingredients.
Where we recommend using the algorithm
Pages
Product card
Cart
Letters
Post-transaction
Accessories
Available solutions:
A variation of the "Related Products" algorithm. Matches additional accessories to the product being viewed. Helps the user to close his need in one place. For example, to pick up a case, headphones and protective glass for a smartphone at once, which increases the average cheque of the online shop.
Unlike the basic algorithm, priority is given to products from categories that we define ourselves based on our experience, as well as products of the same brand.
Where we recommend using the algorithm
Pages
Product card
Cart
Letters
Post-transaction
Related products prioritised by brand of the product being viewed
Available solutions:
A variant of the Related Products algorithm. It is used when there is not enough information about user behaviour.
For example, based on behavioural statistics, we see that a particular chair is bought together with a particular table, then the recommendations will first of all suggest it. But if we don`t know what is usually bought with an Eames dining table, then the algorithm will recommend chairs and other furniture of the same brand.
Where we recommend using the algorithm
Product card
Related products from categories, other than the current product category
Available solutions:
A variant of the "Related Products" algorithm, in which products of the same category as the product being viewed are removed from recommendations.
For example, it makes no sense for an electronics shop to recommend other TV sets to a TV set. But if we are talking about baby food, then apple puree in addition to pear puree would be organic.
Where we recommend using the algorithm
Product card
Cart
Accessories tailored to the shop`s requirements
Available solutions:
Complements the basic "Related Products" algorithm and allows the shop to adjust product rendition using manually configurable rules. It will be useful if you need to promote products from unpopular categories or the shop wants to ensure that only really suitable related products are recommended to the main product. Relevant for technically complex products: snowmobiles, computers and so on.
Where we recommend using the algorithm
Product card
Cart
Accessories tailored to the shop`s requirements
Available solutions:
The algorithm displays up to 20 related categories for a given category, sorted in descending order of popularity. For example, for washing machines, this could be powders, descalers and accessories.
The algorithm is used to simplify navigation: the customer is presented with categories where he can find additional products for his purchase.
Accessories tailored to the shop`s requirements
Available solutions:
A variation of the "Accessories" algorithm with a more diverse output. It is suitable if there are a lot of accessories that are very similar in terms of product properties.
Where we recommend using the algorithm
Product card
Cart
Related products to the last added to basket
Available solutions:
The algorithm is designed for use on product pages. It analyses the contents of the shopping cart and suggests products that can complement the purchase. An attempt to sell related products before the user proceeds to checkout can be more effective than a similar one on the "Basket" page.
Where we recommend using the algorithm
Product card
Ready-made outfits, collages
1 algorithm
Total Look AI Stylist
Total Look AI Stylist
Available solutions:
Some customers find it difficult to choose a harmonious image. This algorithm solves this problem by offering several variants of combining the selected item of clothing with other products available in the shop`s database. This increases not only the probability of buying this item, but also stimulates cross-selling.
For example, if a user is interested in a black turtleneck, the image together with this product will recommend jeans, a jacket, sneakers and a bag of a suitable colour. Images are selected taking into account common stylistic rules, i.e. we will recommend a T-shirt to shorts, but not a warm jacket with a hat.
When building recommendations, the algorithm takes into account the behaviour of other users and the popularity of products.
Where we recommend using the algorithm
Product card
Total Look AI Stylist
Available solutions:
Some customers find it difficult to choose a harmonious image. This algorithm solves this problem by offering several variants of combining the selected item of clothing with other products available in the shop`s database. This increases not only the probability of buying this item, but also stimulates cross-selling.
For example, if a user is interested in a black turtleneck, the image together with this product will recommend jeans, a jacket, sneakers and a bag of a suitable colour. Images are selected taking into account common stylistic rules, i.e. we will recommend a T-shirt to shorts, but not a warm jacket with a hat.
When building recommendations, the algorithm takes into account the behaviour of other users and the popularity of products.
Where we recommend using the algorithm
Product card
Recommendations for the search query
3 algorithms
Basic algorithm
Basic algorithm
Available solutions:
In this scenario, we recommend products that best match the user`s search query. When forming them, the algorithm relies on the behaviour of users who have already searched for something similar. If there are not enough such products, alternatives to them are added.
The algorithm improves the user experience by helping a person find the most appropriate product even for non-standard, complex or error-ridden search phrases. Examples of such queries are: "dishwashing detergent", "bike", "red Delonghi kettle". The algorithm uses classical machine learning and many solutions that have been tested over the years.
Where we recommend using the algorithm
Pages
Search
Empty search
Letters
Abandoned search request
Search recommendations with limiting the influence of super popular products
Search recommendations with limiting the influence of super popular products
Available solutions:
A variant of the "Search Recommendations" scenario, in which the weight of the most popular products is purposely reduced. For example, in order not to recommend bananas in addition to most items in an order in an online grocery shop, the system sort of penalises this product.
Where we recommend using the algorithm
Pages
Search
Empty search
Search recommendations based on neural networks
Search recommendations based on neural networks
Unlike the basic search algorithm, this variant uses deep neural network training: the algorithm "understands" the meaning of the phrase, rather than just pulling out the name of the product, which helps to more accurately determine what the user is looking for.
Basic algorithm
Available solutions:
In this scenario, we recommend products that best match the user`s search query. When forming them, the algorithm relies on the behaviour of users who have already searched for something similar. If there are not enough such products, alternatives to them are added.
The algorithm improves the user experience by helping a person find the most appropriate product even for non-standard, complex or error-ridden search phrases. Examples of such queries are: "dishwashing detergent", "bike", "red Delonghi kettle". The algorithm uses classical machine learning and many solutions that have been tested over the years.
Where we recommend using the algorithm
Pages
Search
Empty search
Letters
Abandoned search request
Search recommendations with limiting the influence of super popular products
Available solutions:
A variant of the "Search Recommendations" scenario, in which the weight of the most popular products is purposely reduced. For example, in order not to recommend bananas in addition to most items in an order in an online grocery shop, the system sort of penalises this product.
Where we recommend using the algorithm
Pages
Search
Empty search
Search recommendations based on neural networks
Unlike the basic search algorithm, this variant uses deep neural network training: the algorithm "understands" the meaning of the phrase, rather than just pulling out the name of the product, which helps to more accurately determine what the user is looking for.
Personalised recommendations
6 algorithms
Basic algorithm
real-time & personal
Basic algorithm
real-time & personal
Available solutions:
The algorithm analyses the user`s behaviour and recommends products that are most interesting to him. If a person has no browsing history yet, popular products can be shown to them.
If the user has shown interest in certain products, the algorithm will select alternative offers and help him find the most suitable one, which will bring him closer to the purchase. If the user has already ordered something, the algorithm will suggest related products.
Where we recommend using the algorithm
404 page
User account
Empty search
Home page
Personalised recommendations category page
real-time & personal
Personalised recommendations category page
real-time & personal
Available solutions:
The algorithm is similar to the basic one, but only products from the current category and its subcategories are recommended. The algorithm uses only the user`s previous actions with products (browsing, adding to basket, orders) and implicit links between products. This algorithm differs from `popular products from categories of interest to the user` in that it analyses short-term interests of the user.
Where we recommend using the algorithm
Category
Products previously viewed by the user
real-time & personal
Products previously viewed by the user
real-time & personal
Available solutions:
The algorithm recommends products that the user has already looked at. This makes it easier to navigate through the shop, as well as reminds the person of their need and encourages them to buy.
Where we recommend using the algorithm
Home page
User account
404 page
Product card (in some cases)
Cart (in some cases)
Personalised recommendations based on past orders
real-time & personal
Personalised recommendations based on past orders
real-time & personal
Available solutions:
The algorithm recommends products that the user has already bought. It takes into account the age and frequency of purchases. The scenario is relevant for online shops that have goods of repeated demand: baby food, pet food, pharmacies, and so on.
Where we recommend using the algorithm
Pages
Home page
404 page
User account
Product card (in some cases)
Cart (in some cases)
Letters
Next Best Offer
Personalising recommendations based on the user`s interest in product features
real-time & personal
Personalising recommendations based on the user`s interest in product features
real-time & personal
Available solutions:
This algorithm works on the basis of the basic one. It generates the output in such a way that at the beginning there are the products that are maximally similar in properties to those that this user was interested in during the last 2 hours. It takes into account brand, price and type of goods.
Where we recommend using the algorithm
Any page, depends on the underlying algorithm
Size-adjusted goods
real-time & personal
Size-adjusted goods
real-time & personal
Available solutions:
Sorts the output of the base algorithm taking into account the size the user is interested in. The information is obtained on the basis of analysis of goods added to the cart. The algorithm takes into account that for different product categories the concept "size" may have its own meaning.
Where we recommend using the algorithm
Home page
Category
Product card
Cart
other
Basic algorithm
real-time & personal
Available solutions:
The algorithm analyses the user`s behaviour and recommends products that are most interesting to him. If a person has no browsing history yet, popular products can be shown to them.
If the user has shown interest in certain products, the algorithm will select alternative offers and help him find the most suitable one, which will bring him closer to the purchase. If the user has already ordered something, the algorithm will suggest related products.
Where we recommend using the algorithm
404 page
User account
Empty search
Home page
Personalised recommendations category page
real-time & personal
Available solutions:
The algorithm is similar to the basic one, but only products from the current category and its subcategories are recommended. The algorithm uses only the user`s previous actions with products (browsing, adding to basket, orders) and implicit links between products. This algorithm differs from `popular products from categories of interest to the user` in that it analyses short-term interests of the user.
Where we recommend using the algorithm
Category
Products previously viewed by the user
real-time & personal
Available solutions:
The algorithm recommends products that the user has already looked at. This makes it easier to navigate through the shop, as well as reminds the person of their need and encourages them to buy.
Where we recommend using the algorithm
Home page
User account
404 page
Product card (in some cases)
Cart (in some cases)
Personalised recommendations based on past orders
real-time & personal
Available solutions:
The algorithm recommends products that the user has already bought. It takes into account the age and frequency of purchases. The scenario is relevant for online shops that have goods of repeated demand: baby food, pet food, pharmacies, and so on.
Where we recommend using the algorithm
Pages
Home page
404 page
User account
Product card (in some cases)
Cart (in some cases)
Letters
Next Best Offer
Personalising recommendations based on the user`s interest in product features
real-time & personal
Available solutions:
This algorithm works on the basis of the basic one. It generates the output in such a way that at the beginning there are the products that are maximally similar in properties to those that this user was interested in during the last 2 hours. It takes into account brand, price and type of goods.
Where we recommend using the algorithm
Any page, depends on the underlying algorithm
Size-adjusted goods
real-time & personal
Available solutions:
Sorts the output of the base algorithm taking into account the size the user is interested in. The information is obtained on the basis of analysis of goods added to the cart. The algorithm takes into account that for different product categories the concept "size" may have its own meaning.
Where we recommend using the algorithm
Home page
Category
Product card
Cart
other
Find out how the recommendations will help your business
We will show the work of all recommendation algorithms, tell you about the Retail Rocket Group ecosystem and answer your questions.
Error: Contact form not found.