Skip to main content

Marketplace Recommendation Recipes

Recipes for multi-seller platforms, focused on seller-based recommendations and time-sensitive listings. Featured recipes (☆) are listed first. Expand a recipe to see its use cases, requirements and how it works.

Recipes for other industries: Cross-industry · Retail · Media · Automotive · Online Learning · Location-Based Businesses

New users

For users who have recently joined or started engaging.

Top wishlisted items based on input category and gender
Top wishlisted items based on input category and gender with size…

Use cases

Use provided input category to provided relevant rec­om­men­da­tions by gender.

Re­quire­ments

  • User gender.

  • Catalog with category and size (where applicable).

  • Event activity to indicate popularity.

Inputs

  1. Input category.

  2. Gender category ranking to use.

  3. Category ranking to use if gender unavailable.

  4. Should size filter be applied?

  5. If yes, provide activity that should be used to determine size.

How it works

Based on user gender and input category, show top wishlisted items.

Active users

For users who engage regularly.

☆ Featured · Next Best Items
Post activity follow up alerts

Use cases

  • Post activity follow up alerts.

  • Abandoned search/​browse/​cart follow ups.

Re­quire­ments

  • View or equivalent events with product_ids.

  • Item catalogs with matching product_ids and size in tags ( as applicable).

Inputs

  1. Recent activity timeline.

  2. Precomputed rankings based on other users with similar activity => Next best product, Cross-sell products.

  3. Category backfill.

  4. Site backfill.

How it works

Top items to consider after recent activity based on users with similar history and custom filters.

Feed type

Product-Product

☆ Featured · Store name affinity items
Popular, Trending or Recent Items by user affinity store names

Use cases

  • Daily/Weekly newsletter with top/trending items in user affinity store names.

  • App/Site live content with top/trending items in user affinity store names.

Re­quire­ments

  • Item catalogs with store name.

  • Items activity/event timeline.

Inputs

  1. Number of items to recommend per affinity store name (Default 2).

  2. Store Name Ranking.

  3. Site-wide Backfill (if rec­om­mend­ed products less than return count). Set 'None' if not required.

How it works

Popular, Trending or Recent Items by user affinity store names.

Daily sponsored feed
User Per­son­al­ized Feed like Items For You / Top Picks for You

Use cases

  • User Per­son­al­ized Feed like Items For You / Top Picks for You.

  • Daily/Weekly Newsletters.

  • Homepage recommendations.

Re­quire­ments

  • Category in product catalog.

  • User event timeline activity.

  • sponsored attribute on catalog.

  • Tags on catalog.

Inputs

  1. User Recent Activity.

  2. Consider items from category.

  3. Use tag similarity to recommend products based on recent activity.

  4. Provide next best product ranking to use for recent activity 2. Site-wide backfill (optional).

How it works

Per­son­al­ized list of items generated from user recent activity that are from com­bi­na­tion of items from tag similarity, next best product and category trending products that are sponsored.

Live content - top wishlisted items by gender category and size filter
Top wishlisted items in live content payload and affinity categories…

Use cases

Use on webpage or app to show relevant rec­om­men­da­tions to user based on current browse, past affinity and gender filtered by size (as applicable).

Re­quire­ments

  • User gender (on user profile or in live content call payload).

  • Catalog with category and size (where applicable)).

  • Event activity to indicate user affinity, top wishlisted items etc.

Inputs

  1. Should user affinity categories be used along with payload categories?

  2. Number of items per category.

  3. Gender category and category rankings to use.

  4. Should size filter be applied?

  5. If yes, provide activity to determine size filter.

How it works

Based on user gender, user affinity categories and seed categories provided as part of live content payload, show top wishlisted items including optional filtering on size Live content payload sample - There must be atleast 1 seed_item_id or seed_category to drive the recommendation.

{ "page_number": 1, "seed_item_ids": '592189667' , "seed_categories": "Accessoarer" , "exclude_item_ids": '592189667' , "exclude_categories": , "gender": "male" }

Similar items to previous block (product + seller_id) with size filter
Use products in previous blocks to recommend similar items

Use cases

  • Use products in previous blocks to recommend similar items.

  • Multi block per­son­al­i­za­tion based on recent activity.

Re­quire­ments

  • Events with product_ids.

  • Item catalogs with matching product_ids and seller_ids and size (where applicable).

Inputs

  1. Previous block name.

  2. Number of items to consider from recent activity. Default value is 1.

  3. Seller backfill for given seller_id level. Select None, is not required.

  4. Site backfill. Select None, is not required.

  5. Apply size filter?

  6. If yes, provide activity to determine size filter.

How it works

Items related to previous block with backfills and optional size filtering. Similarity is based on user session and trans­ac­tion activity with other users. This is backfilled with other products based on the seller_id of previous block products.

Intermittently active users

For users who engage now and then.

☆ Featured · New/trending from your favorite sellers
Re-engage users based on past activity

Use cases

  • Re-engage users based on past activity.

  • Expand item discovery based on users activity.

  • Daily/Weekly customised feed to users.

Re­quire­ments

  • Item catalogs with seller.

  • User recent activity timeline.

Inputs

  1. Recent event timeline.

  2. Seller top items ranking to use for each seller.

  3. Site level backfill ranking (if rec­om­mend­ed products less than return count). Set 'None' if not required.

  4. Apply size filter?

  5. If yes, provide recent activity to use for determining size filter.

How it works

Based on recent activity sellers, recommend top ranking products from the sellers with size filter.

Churned users

For users who have stopped engaging.

☆ Featured · Ending soon items
Alert users to expiring items and promote related discounts

Use cases

  • Alert users to expiring items and promote related discounts.

  • Daily or Weekly Newsletters.

  • Backfill for other recommendations.

Re­quire­ments

  • Catalog of items.

  • end_date attribute in ISO8601 format.

Inputs

Site wide pre-ranked list of items sorted by end_date timestamp.

How it works

Items in catalog with end_date less than next few days indicating expiry or removal.

Affinity Category tags Items
Items matching category and tags that user has affinity towards

Use cases

  • Daily/Weekly newsletter with top/trending items in user affinity categories tags.

  • App/Site live content with top/trending items in user affinity categories tags.

Re­quire­ments

  • Item catalogs with category and tags.

  • User activity timeline with events.

Inputs

  1. Number of items to recommend per affinity category and tag (Default 2).

  2. Items ranking to fetch candidate at category and tags.

  3. Site level backfill ranking (if rec­om­mend­ed products less than return count). Set 'None' if not required.

How it works

Popular, Trending, Recent or Deal Items in user affinity categories and tags.

Items matching User Seller Attributes
Popular items cor­re­spond­ing to user selected sellers

Use cases

  • Popular / Trending items per­son­al­i­za­tion based on specific user attribute.

  • Mix and match with pre-computed user rankings.

  • Daily / Weekly Newsletters.

  • User specific promotions / sponsored items.

Re­quire­ments

User attribute value matching seller values from product catalog.

Inputs

  1. User Attribute Name.

  2. Seller Ranking.

  3. Site-wide backfill (optional).

How it works

Popular items from user favorite sellers joined with item catalogs. The list of sellers must match what is available in catalogs and are case sensitive.

All users

For users at any lifecycle stage.

Live content - top wishlisted items by seed sellers
Based on seed_seller_ids provided, return top x products by seller.…

Use cases

Use on webpage or app to show relevant rec­om­men­da­tions to user based on current browse, past affinity => Various selections of live content payload and affinities.

Re­quire­ments

  • Catalog with category, tags, seller_id.

  • Event activity to indicate user affinity, top wishlisted items etc.

Inputs

  1. Live content payload with seed_​seller_​ids.

  2. Supplement with user favorite sellers?

  3. Seller ranking.

  4. Apply size filter?

  5. If yes, provide activity to use to determine size filter.

How it works

Based on seed_seller_ids provided, return top x products by seller. If affinity seller option selected, supplement with user favorite sellers.

Live content - top wishlisted items by seed sellers & gender
Based on seed_seller_ids and gender provided, return top x products…

Use cases

Use on webpage or app to show relevant rec­om­men­da­tions to user based on current browse, past affinity => Various selections of live content payload and affinities.

Re­quire­ments

  • Catalog with category, tags, seller_id.

  • Event activity to indicate user affinity, top wishlisted items etc.

Inputs

  1. Live content payload with seed_​seller_​ids.

  2. Supplement with user favorite sellers?

  3. Seller ranking.

  4. Apply size filter?

  5. If yes, provide activity to use to determine size filter.

How it works

Based on seed_seller_ids and gender provided, return top x products by seller. If affinity seller option selected, supplement with user favorite sellers.

Live content - site items ending soon / top x
This can be used for anonymous users or new users where sufficient…

Use cases

Use on webpage or app to show ending soon items or top wishlisted items for anonymous or new users.

Re­quire­ments

  • Catalog with end date.

  • Event activity to indicate top items.

Inputs

  1. Optional live content payload with page number, exclude_​item_​ids and exclude_​categories.

  2. Choose - ending soon or top wishlisted => The recipe uses a com­bi­na­tion of rankings to create an effective candidate set.

  3. Apply size filter?

  4. User activity to use to determine size filter.

How it works

This can be used for anonymous users or new users where sufficient Depending on choice - top wishlisted or ending soon, top items at site level are selected and filtered per choice. Sensitive content is filtered out.

User Coupons matching brand in event / live content payload
User coupons based on brand provided in event or in live content…

Use cases

Website/App per­son­al­ized offers based on user search on specific brands.

Re­quire­ments

  • Live content call/ event call with brand(s) in payload.

  • Item catalogs matching the seed brand(s).

Inputs

  1. Brand attribute name in live content payload.

  2. Brand ranking.

How it works

User Coupons matching brand in event / live content payload.

Connect with Blueshift for personalized recommendations

Whether you're exploring new ways to use recommendation recipes or looking to design a custom model for your business, our team can guide you through the next steps.

Reach out to [email protected] or your Customer Success Manager (CSM) for assistance.

What's next

Learn how Recommendations Studio works and understand the workflow for creating personalized recommendations.
​How Recommendations Studio works →

Related Articles
Did this answer your question?