Product discovery

This guide covers the most common use cases connected to product discovery. Use it to find the use case that best suits your goal and follow the setup guidance to get it running.

Prerequisites

Use cases in this guide require catalog attributes or events to be mapped before they work correctly. See data mapping for guidance on mapping catalog attributes and events to the naming conventions Bloomreach expects.

New products on stock

Ecommerce goal: Activation, repeated purchases.

Example usage: Recommending similar items engages customers with a variety of products related to the one they're currently viewing or recently viewed. When your catalog is well-documented and rich in properties, you can surface alternatives based on shared product attributes.

Ideal placements: Homepage, campaigns.

Requirements: A product catalog containing a datetime column or any numerical column used to sort products in descending order.

Setup: Use the New items template.

Alternative products (based on items)

Ecommerce Goal: Activation, repeated purchases.

Example usage: Recommending similar items engages customers with a variety of products related to the one they're currently viewing or recently viewed. When your catalog is well-documented and rich in properties, you can surface alternatives based on shared product attributes.

Ideal placements: Product detail page.

Requirements: A product catalog containing as many features as possible, such as descriptions, images, tags, brands, categories, and colors. To make these attributes available for recommendations, complete the data mapping.

Setup: Use the Textual similarity model or the More like this model, as used in the Alternative products use case.

Alternative products (based on customers)

Ecommerce Goal: Activation, repeated purchases.

Example usage: When your catalog isn't well-documented — for example, it lacks tags or other grouping properties — you can recommend alternative products based on customer behavioral data instead.

Ideal placements: Product detail page.

Requirements:

• Event tracking for the following events: purchase product, view product, and add to cart.
• At least 3 months of tracked data.

Setup: Use the Product detail template. You can also set customer preferences reordering. See Alternative products (based on items) for an example.

Alternative products (based on filter)

Ecommerce goal: Activation, repeated purchases.

Example usage: Recommending products based on catalog filters — such as a particular category, brand, or combination of filters — lets you surface specific products to customers you want to target.

Ideal placements: Product listing page.

Requirements: A product catalog containing as many features as possible, such as descriptions, tags, brands, categories, and colors. To make these attributes available for recommendations, complete the data mapping (link).

Setup: Use the Filter based template.

Most popular products

Ecommerce goal: Activation, repeated purchases.

Example usage: Recommending the most popular items on the homepage is a standard way to engage new customers who have no historical data. Pair it with a prominent header for best results.

Ideal placements: Homepage, email campaigns.

Requirements:

• Data mapping for events that express a popularity metric, most commonly view_item or purchase_item.
• At least one week of tracked data.

Setup: Use the Popular right now template.

Top products from category

Ecommerce goal: Activation, repeated purchases.

Example usage: Showing customers the top products from the category they're currently viewing helps surface relevant items without leaving the page.

Ideal placements: Category page.

Requirements:

• Data mapping for events that record customer interactions with items and represent similarities, such as view_item and purchase_item.
• Category tracking in events. To generate recommendations for categories such as t-shirts, jeans, or skirts, you must track those categories in the relevant events:
  • For top-selling items from a category, track the category in the purchase_item event.
  • For most-viewed items, track the category in the view_item event.
  • Categories can be tracked as a string or a list of strings — for example, category: "t-shirt" or category: ["t-shirt", "woman", "blue"]. If you track categories as a list, verify the format in Data & Assets > Data manager > Events > view_item. The same format applies to one-time imports: use ["t-shirt", "woman", "blue"] with type list for multiple categories, or "t-shirt" with type string for a single category.

Setup: Use the Metric based category template.

Browse abandonment

Ecommerce goal: Repeated purchases.

Example usage: Targeting customers who recently left your website and reminding them of the products they interacted with — based on views or their cart — is an effective approach. Email campaigns featuring the last items a customer viewed work well for this.

Ideal placements: Email, homepage, product detail page.

Requirements: Data mapping for events that record customer interactions with items, such as view_item and purchase_item.

Setup: Use the Customer recent interactions template.