Product Recommendations are commonly utilized to aid in product discovery and drive conversion rates by increasing relevance, boosting average order value through up- and cross-selling, facilitating personalized recommendations for increased purchases, and enhancing merchandising with effective filtering.

For additional commercial examples, strategies, and tactics, please reach out to your Customer Success Manager, contact Nosto support, or visit the Customer Hub.

The basic options for recommendation are

The page type option determines where the recommendation is displayed in the analytics table, along with suggested recommendation types and titles. Choose the most suitable option that describes the target page.

Placements are Nosto's product concept used to populate Onsite Product Recommendations, Onsite Content Personalization or Visual UGC campaigns, and to set up Nosto's A/B Tests on the target website. Placements define where Nosto campaigns are displayed on the website, allowing further customization based on the page's or user's context. Learn more about placements and their use cases here.

Segments are Nosto's product concept for the intended target audience. They enable targeting broad audiences from the 'everyone' segment, referring to all website traffic, to macro-audiences such as 'mobile' and 'desktop shoppers', and even custom micro-audiences created under Segmentation & Insights. In simple terms, placements determine where recommendations are displayed, while segments define who sees the recommendations. Learn more about Segmentation here.

As the name suggests, Scheduling allows you to time the activation and termination of recommendations and other Nosto features on specific dates, times, or weekdays.

Learn more about Scheduling here.

The recommendation type is the algorithm that ultimately controls the output of the recommendations, specifically the recommended products. The algorithm's output can be further controlled and customized using options available later in the workflow. The Recommended tab displays suggested types based on the chosen page type, while the All types tab lists all available recommendation types. Learn more about recommendation types here.

Each recommendation type has its own set of type-specific settings, ranging from Time Periods considered for Best Sellers to Visibility Settings for Free Shipping Recommendations.

For instance, Best Sellers can be based on orders within the past 24 hours or a longer time period, while Free Shipping Recommendations offer options to choose the cart value criteria that activate or hide the recommendation. Detailed explanations of the available options are provided here.

Filters provide an additional layer of customization for recommendations, allowing the algorithm to narrow down the scope of recommended products based on specific criteria.

Basic filters consist of inclusive and exclusive rules that restrict the displayed products based on given criteria. For instance, cross-sellers can be limited to products from the same product range as the viewed one by applying a price filter to remove or reduce cheaper alternatives.

After applying filters, the algorithm can fill any available or empty item slots based on the given criteria. For example, if the maximum amount of products was set to ten (10) in the initial setup, but the filtered algorithm only yields eight (8) products, the remaining two (2) item slots can be filled with similar unfiltered options or only filtered options, depending on the chosen setting. Learn more about filters here.

Advanced Dynamic Filtering enables the creation of precise recommendation pairings or merchandising by narrowing down recommendations to hand-selected product subsets based on specific product attributes. This feature, primarily designed for product detail pages, utilizes behavioral data similar to regular filtering but applies stricter rules to showcase the best and most relevant options.

Attributes category, brand, tags, product ID, and product price can be used in Advanced Dynamic Filtering. It is represented by if-this-then-that rules. It's important to note that this feature differs from Dynamic Bundles, which supports more advanced use cases and a wider range of tools.Learn more about advanced dynamic filtering here.

Variant settings serve as a personalization layer that further narrows down recommended products based on the variant type that aligns with the shopper's indicated interest or known preference by excluding variants which aren’t available.

Variants can represent different types a product is sold in, such as size and color for garments or bottle size in cosmetics. Size and color are commonly used variant types and serve as examples.

When the Nosto integration is configured to feature product variant data and Nosto processes shopper's variant signals, variant settings act as the final layer that restricts the output of recommended products based on the configured preferences in previous steps. Variant settings offer two distinct modes: Direct Interactions and Known Affinities, each suited for different scenarios and use cases.

Learn more about variants here

Visual settings encompass the options for the visual representation of the recommendation.

The title serves as the visible header for the recommendation, which shoppers will see on the website.

Titles can also be generated, suggested, and translated using the GenerativeAI Copy Creation Tool. If the title field is left empty, GenerativeAI suggests titles based on the known attributes of the recommendation defined in previous steps, and taking into account configurable options such as creativity and business vertical. If a title has been defined, the tool suggests alternative expressions based on the provided title.

The translation mode offers suggestions for translating the titles to the target language, either based on the input or independently. It is advisable to review translated titles if you are not proficient in the target language, although the quality is generally good even for less commonly spoken languages.

Templates are web design templates comprising HTML, .css, and possibly JavaScript. They define how the recommendations will appear on a website. Templates do not include specific content such as products but determine the rendering of recommendations in a browser. Templates are primarily constructed and adjusted by web designers and are typically built once or whenever site design changes. A template is attached to a recommendation created or edited in the workflow, and it then loads in the defined placement.

Please note that your implementation method may differ. For example, if the website is built using JavaScript frameworks like AngularJS or React, it is recommended to adhere to the guidelines provided here or contact Nosto’s support.

Learn more about HTML-based template building here.

The main option for fallbacks generates product recommendations when the primary recommendation is not applicable or lacks sufficient data to meet the criteria.

A practical example is configuring the primary recommendation to display highly personalized recommendations based on shopper’s browsing and shopping history. However, for completely new shoppers without any browsing history, a fallback recommendation can be used to feature best-selling products as a generic recommendation, which replaces recommendation making use of browsing.

Fallbacks provide a simpler method for this use case, although achieving the same outcome can also be done using segments e.g. new visitors; anyone returning to the website segments.

Fallback options offer the same configurations as the primary recommendation, inheriting its styling.

Behaviour option determines what happens when the minimum product criteria defined in the initial step for the primary recommendation cannot be fully met.

The summary provides an overview of the recommendation configuration. Clicking "Save" will immediately enable the recommendation. Recommendations can also be saved as drafts by leaving them in an inactive state in the first step and by saving them.

It's important to note that recommendation creation and editing do not enforce the following:

All of these factors can keep the recommendation hidden or inactive.