An overview of Dynamic Yield’s Essential Recommendation Capabilities
Blog

An overview of Dynamic Yield’s Essential Recommendation Capabilities

After spending nearly a decade developing the most powerful recommendation engine in the market, take a look at an overview of the current recommendation capabilities available to Dynamic Yield customers.

In today’s hyper-competitive retail environment, consumers expect all of their digital experiences to be personalized and relevant, including the recommendations they receive from eCommerce companies. Machine learning has become the crux of personalizing the recommendation experience for visitors, assisting in making generalizations from observed data to better find the best products or offers to serve each specific shopper.

We have spent nearly a decade developing the most powerful recommendation engine in the market, built based on direct feedback from thousands of merchandisers and retailers across hundreds of global brands and verticals.

With Dynamic Yield, you can exercise full merchandising control over your content or product recommendations; adapt the recommendation layout according to context, fuse multiple strategies into a single recommendation widget, test and optimize any element – including the layout, design, recommendation strategy, location, widget headers, and more.

You can also predict customers’ next purchase through a self-training, deep learning recommendation strategy to automatically predict the next product a user may be interested in by exploiting the co-occurrence of products in a user’s browsing history.

This in-depth overview includes a comprehensive look at all of our existing recommendations capabilities, broken down across 13 categories.

Widget Creation
(UX & Design)

1. Flexible Design & Customization

Developers can build and design templates that are flexible for marketers to re-use, alter and scale across the site.

2. Recommendation Widget Preview

When the widget is rendered - this provides the ability to see the recommended results on the page without publishing the experience.

3. Out-of-the-box Widget Templates

Enables fast implementation of ready made designs.

4. Templating Engine

Allows saving recommendation widget design as templates - to be reused across multiple widgets.

5. Template History

View the history of changes made to the code of the templates. This can be used to trace the source of a problem (troubleshooting), or retrieve a previous version of your code.

Targeting
Conditions

6. Time-Based Targeting

Schedule when experiences should be live and removed from site.

7. Affinity-Based Targeting

Target users who expressed high interest - measured by their interactions with products (views, add to cart and purchase events) - in a specific product property (e.g. blue color, show category).

8. New User (First Session) Targeting

Target users who are currently in their first session ever (since Dynamic Yield script was implemented to the site).

9. Session Behavior Targeting

Enables targeting the recommendation widget experience by session behavioral properties like number of page views, current URL, previous URL, elapsed time, etc.

10. Shared Audience-Based Targeting

Target users who belong to a Dynamic Yield audience from a different Dynamic Yield site.

11. Traffic Source-Based Targeting

Enables targeting for users arriving via a specific traffic source.

12. Audience-Based Targeting

Target users who belong to a Dynamic Yield audience from this Dynamic Yield site.

13. Technology-Based Targeting

Enables targeting the recommendation widget experience by device, browser, operation system, operating brand, screen resolution, and user agent.

14. Location-Based Targeting

Enables targeting the recommendation widget experience by the user’s current location.

15. Weather-Based Targeting

Enables targeting the recommendation widget experience by the current or forecasted weather in the user’s current location.

16. Product Engagement / Social Proof

Enables targeting the recommendation widget experience to product pages displaying products that have received a given amount of views and purchases.

17. Page Property Targeting

Enables targeting the recommendation widget experience to product pages displaying products that have received a given amount of views and purchases.

18. Custom Evaluators

Enables targeting the recommendation widget experience by any property calculated adhoc using custom JS code.

Recommendation
Strategies

19. Automatic Recommendation Strategy

This strategy selects the best algorithm based on best practices for the page type chosen.

20. Affinity-Based Personalized Recommendations

Selects the most relevant products by all product engagements tracked for the user. Considers engagements from 2 years back up to engagements performed in the current session (real-time).

21. Purchased Together Recommendations

Selects products frequently purchased with the product being viewed or the set of products in the user’s cart. Can consider purchases made online or offline.

22. Viewed Together Recommendations

Selects products frequently viewed together with the product being viewed in the same session.

23. Recently Viewed-based Recommendations

Recommends the last items viewed by the current user (most recent appears first).

24. Recently Purchased-based Recommendations

Recommends the last items purchased by the current user (most recent appears first).

25. Most Popular Recommendations

Scores items based on the weighted sum of all interactions – such as purchase, add to cart and product view – favoring recent interactions over historical ones.

26. Most Popular in Category

Returns items by popularity, but only includes items in the category specified in the page context of the category page.

27. Similar Items

Recommends items that are similar to the item currently displayed, factoring in item popularity.

28. Personalized Recommendations (Collaborative Filtering)

Selects the most relevant products by comparing similar behaviors across all users using sophisticated machine learning models.

29. Purchased Together Offline Recommendations

Recommends products that have been purchased offline together with the item currently displayed or the set of products in the user’s cart.

30. Purchased Together Offline & Online Recommendations

Recommends products that have been purchased together offline or online with the item currently displayed or the set of products in the user’s cart.

31. Purchased with Recent Purchased Recommendations

Items which are usually purchased together with the last items purchased by the current user.

32. Last Purchased Recommendations

Presents the cart content of the most recent purchase by the current user.

33. Purchased with Last Purchased Recommendations

Items which are usually purchased together with items in the most recent purchase by the current user.

34. Fallback Strategies

In case the strategy returns fewer items than number of slots, the system will run a respective fallback algorithm for each strategy.

35. Deep Learning Recommendations

A self-training deep learning recommendation strategy to automatically predict the next product a user may be interested in based on the sequence of both historical and real-time interactions.

Dynamic
Strategy Filters

36. Product Properties

Setting allows the option for recommendations to show or exclude products that have all or a selected amount of product properties as the product being viewed. This capability is limited to the product page only.

37. Exclude recently viewed products

Removes recently viewed products from being displayed.

38. Exclude recently purchased products

Removes recently purchase products from being displayed.

Recommendation
Strategy Management

39. Shuffle Recommendation Results

This strategy selects the best algorithm based on best practices for the page type chosen.

40. Mixed Recommendation Strategies (Algorithmic Fusion)

Enables setting different recommendation algorithms per widget slot.

Merchandizing
Rules & Control

41. Real-Time Filters with Recommendation API

Add real-time filter rules to filter the results based on data obtained within the session (e.g. show products priced higher than the currently viewed product, present products based on the visitors explicit selection, etc.).

42. Custom Filtering Rule Creation

Allows setting rules that restrict the recommendations results.

43. Pin Products

Allows the user to force displaying specific products.

44. Whitelist Products (Include)

Ensure that only a subset of products will be presented in a specific widget or a specific slot.

45. Blacklist Products (Exclude)

Allows the user to restrict what products should not be recommended in a specific widget or a specific slot.

46. Dynamic List Creation

Allows selecting products to pin / whitelist / blacklist using a set of product attribute conditions (e.g. price is higher than $100, category is ‘Dresses’, etc.).

47. Merchandizing Rule Prioritization

Allows resolving conflicts between 2 contradicting rules.

48. Merchandizing Rule Targeting

Allows setting the conditions in which a rule should be applied (to what users, in what pages, etc.).

49. Merchandizing Rule Scheduling

Allows setting the date and time in which a rule should be applied.

Testing &
Optimization

50. Recommendation Algorithms Testing

A/B test different recommendation algorithms.

51. Recommendation Widget Layouts Testing

A/B test different recommendation widget layout and design (vertical vs horizontal, number of slots, showing/hiding the price, etc.)

52. Merchandizing Merchandizing Rule Testing

A/B test different recommendation filtering rules.

Channels

53. Web & Mobile Web

Renders recommendations anywhere on the web and mobile web site.

54. Mobile App

Renders recommendations anywhere on the mobile app.

55. Email

Renders recommendations on emails.

Reporting

56. Primary Metric Reporting

Renders recommendations anywhere on the web and mobile web site.

57. Enhanced Experience Reports – Secondary Metrics

Displays how variations are affected by metrics that are not the primary metric of the test. The available secondary metrics are CTR (in campaigns types that inject HTML), page views, Purchases, Revenue, AOV, and any custom event.

58. Probability To Be Best

In A/B testing - indicates each test variation’s probability to be the test’s winning variation (in terms of a specific primary metric. e.g. purchase revenue) based on DY’s Bayesian statistics model.

59. Enhanced Experience Reports – Audience Breakdowns

Allows comparing the performance of a recommendation widget between different audience groups.

60. Custom Detailed Recommendations Reports

Displays fine granular reports for recommendation widgets. (i.e show the direct & assisted impact of product recommendations on site’s performance with the ability to have flexible attribution windows).

61. Outlier Filtering

Prevents individual events from skewing results and incorrectly predicting average future user behavior.

62. Predictive Targeting

Automatically suggests what recommendation strategy should be applied per audience based on predicted performance.

Dynamic
Recommendations

63. Open-Time Rendering (Email Recommendations)

Recommendations are rendered at open-time to ensure algorithm calculations and product properties are up-to-date.

64. Triggered Emails

Sends an email with featured product(s) to users who have left site (or app) without completing a purchase. The email includes content relevant to the abandoned browsing session.

65. Push Notifications

Sends a push notification comprised of dynamically selected product(s) and (optional) action buttons at a schedule time or if an event is triggered.

66. Ads

Replaces static Display Recommendation Ads with recommendations driven by machine learning algorithms that choose the best variation in real time.

67. Landing Pages

Delivers personalized recommendation variations for each user instead of a single variation (one-size-fits-all).

Integrations

68. Product Feed Support

Contains the entire product catalog (products, with product metadata) and power product recommendations. Can fully ingest and handle large feeds with millions of product SKUs to power recommendations.

69. Delta Feed API

Sends real-time changes of items to the product feed.

70. API Control

Allows receiving recommendation results in raw format - to leverage recommendations in any external system and to render using custom code, using both server- or client-side deployments.

71. Contextual Widget Placement

Allows configuring where a recommendation widget should be placed without any coding.

72. Multi-Feeds Support

Automatically combine multiple files or sources into one feed.

73. Mutli-Language Campaign Support

Allows multiple values for the same item in a data feed that represent different languages such as multiple product names and descriptions

Testing
Types

74. Manual Traffic Allocation (Standard A/B Tests)

Allows to test recommendation variations based on the defined percentage of traffic allocation.

75. Automated Traffic Allocation (Multi-Armed Bandit)

Automatically decides what percentage of site users will receive each variation. It leverages a Multi-Arm Bandit strategy, which means that variations that performed better in the past are allocated more traffic.

Recommendation
Types

76. Product/Item Recommendations

Displays the most relevant products/items for each user based on the context and goal of the recommendation widget.

77. Content Recommendations

Displays the most relevant content for each user based on the context and goal of the recommendation widget.

Sign up for the XP² newsletter

Join thousands of readers from Target, Citi, Spotify, Hulu, Google, Sephora, and other innovative brands who read our bi-weekly XP² newsletter, delivering educational content, research, and insights straight to your inbox
You may unsubscribe at any time. Visit our privacy policy to learn more about how we process your data and your rights as a data subject.

Thanks for
signing up!