Recommendation Engines let you define which products Merchandising Cloud recommends to shoppers in different areas of your storefront. Each engine combines a Page Type with a Recommendation Type, then applies display settings, fallback behaviour, and optional filters to control the products that can appear.
You can create multiple Recommendation Engines for different storefront contexts. For example, a homepage engine can surface popular or newly added products, while a product-page engine can use the product currently being viewed to recommend related, complementary, or visually similar products.
To access
To access Recommendation Engines:
- From the navigation menu, select Recommendations.
- Select Recommendation Engines.
The Recommendation Engines page lists the engines configured for the selected store. From here, you can review existing Recommendation Engines or select New Engine to create one.
Each entry shows the engine’s name, the page context it is associated with, and the underlying Recommendation Type. This gives you a quick overview of how recommendations are currently configured across the storefront and helps you identify the engine you want to review or update.
Create a Recommendation Engine
To create an engine:
- From the Recommendation Engines page, select New Engine.
- Enter a Recommendation Name.
The Recommendation Name is used to identify the engine in Merchandising Cloud. It does not need to match the underlying Recommendation Type.
For example, you could name an engine Visually Similar while configuring it to use the Visual Recommendations Recommendation Type.
You can now move through the creation flow to define where the engine will be used and how its recommendations should be generated.
Select a Page Type
Select the storefront context where you intend to use the Recommendation Engine. This tells Merchandising Cloud what type of page the recommendation will appear on and helps determine which Recommendation Types are relevant for that context.
For example, a Home Page can use broadly applicable recommendations such as popular or newly added products. A Product Page can also use recommendation types that depend on the product currently being viewed, allowing the recommendation to use that product as context.
Available Page Types include:
- Home Page
- Category Page
- Product Page
- Shopping Cart Page
- Custom Page
Select the required Page Type, then select Next.
The Recommendation Types available in the next step can change depending on the Page Type you select.
Select a Recommendation Type
The Recommendation Type determines how Merchandising Cloud selects products for the engine. The options shown on this page depend on the Page Type selected in the previous step.
When Product Page is selected, additional Recommendation Types become available because Merchandising Cloud can use the product currently being viewed as context.
1. Popular products
Popular products surfaces products receiving strong recent engagement. The recommendation considers product views and purchases, with purchases treated as a stronger indication of popularity.
Because this Recommendation Type does not depend on a particular product being viewed, it can be used for broader product discovery.
2. You recently viewed...
You recently viewed... helps shoppers return to products they have previously explored. It can display up to 20 products viewed by the shopper during the previous 30 days and updates as additional products are browsed.
The products shown therefore depend on the individual shopper's browsing history.
3. Newest products
Newest products surfaces the most recently added products in your catalogue. Products are ordered using their created_at timestamp, from newest to oldest.
This Recommendation Type does not depend on shopper behaviour or engagement history, allowing new products to appear before they accumulate views or purchases.
4. Personalized recommendations
Personalized recommendations uses the shopper's browsing history to recommend products related to their interests. As shoppers interact with products, Merchandising Cloud builds a profile that gives greater importance to recent views and continues to adapt as their interests change.
Recently viewed products are excluded so that the engine can introduce other relevant products. The shopper must have visited at least one product page before there is browsing history available to generate a personalized recommendation.
5. Trending Products
Trending Products identifies products whose engagement is gaining momentum. Merchandising Cloud analyses changes in product views over a two-week period, with greater importance placed on recent activity.
The recommendation considers both growth in views and recency-weighted popularity, helping distinguish products that are actively trending from products that simply have a high overall view count.
6. Customers who viewed this also viewed...
Customers who viewed this also viewed... recommends products shoppers commonly explore during the same browsing sessions as the product currently being viewed. Merchandising Cloud learns these relationships from co-view behaviour over time.
This allows the engine to identify relevant product relationships based on actual shopper behaviour rather than relying only on catalogue attributes.
7. Frequently purchased together
Frequently purchased together recommends products that repeatedly appear together in orders. Merchandising Cloud analyses purchase history and ranks product relationships according to how frequently the products are purchased together.
Product pairs with insufficient co-purchase data are filtered out so that the engine uses stronger purchase relationships when returning results.
8. Customers who viewed this ultimately bought...
Customers who viewed this ultimately bought... connects product views with later purchase behaviour. Instead of looking only at products purchased together, the engine identifies products shoppers eventually purchased after viewing the current product.
These relationships are directional, and the current product is excluded from recommending itself. Low-confidence relationships are also filtered out.
9. Visual Recommendations
Visual Recommendations finds products with visual characteristics similar to the product currently being viewed. In the current interface, this type is available when creating a Recommendation Engine for a Product Page.
The Recommendation Engine can still have its own descriptive name. For example, engines named Visually Similar and Similar Looking Products can both use Visual Recommendations as their Recommendation Type.
Select the required Recommendation Type, then select Next.
Configure the Recommendation Engine settings
After selecting the Recommendation Type, the Settings page lets you control the number of products returned and configure what should happen if the primary recommendation cannot be populated.
Set the number of products displayed
Under General Settings, configure Number of products displayed.
Setting | What it controls |
|---|
Min | The minimum number of products the recommendation should display. |
Max | The maximum number of products the recommendation can display. |
Set these values according to the number of products you want the storefront recommendation placement to support.
Configure Fallback Settings
Some Recommendation Types rely on sufficient shopper, product, or behavioural data before they can return recommendations. If the original engine cannot be populated, a fallback lets Merchandising Cloud use another Recommendation Engine instead.
For example, you can use a more broadly available engine such as Popular Products or Newest Products as a fallback for a recommendation that depends on more specific behavioural data.
To configure a fallback:
- Go to Fallback Settings.
- Select Add Fallback.
- Open Select Fallback.
- Select the Recommendation Engine you want to use.
Only active Recommendation Engines are available for selection as fallbacks.
This means the primary engine is used when it can generate the recommendation, while the selected fallback provides an alternative when the primary engine cannot be populated.
Add Recommendation Notes
Use Recommendation Notes to record additional information about the purpose or intended use of the Recommendation Engine. This can make it easier to distinguish between engines when your store has multiple recommendations configured for different pages or use cases.
Recommendation Notes are for managing the engine and do not change how products are selected.
Configure Filters
After configuring the Recommendation Engine settings, continue to Filters.
Filters are optional rules that give you additional control over the products considered by the engine. They can be used to determine which products should be included or excluded, or to influence how matching products are positioned within the recommendation.
A filter is configured in the following order:
Action → Field → Type → Value
Choose an Action
The Action determines how products matching the rule should be treated.
Available actions include:
Action | What it does |
|---|
Include | Includes products matching the configured criteria. |
Exclude | Excludes products matching the configured criteria. |
Boost | Gives matching products greater prominence in the recommendation. |
Bury | Reduces the prominence of matching products. |
Pin | Pins matching products according to the configured rule. |
Choose a Field
After selecting the Action, choose the product field the rule should evaluate.
The fields available here depend on the product and catalogue data configured for your store. Depending on your setup, you may see fields such as SKU, category, tags, colour, size, price, or custom product fields.
Because these fields come from your store data, the options can differ between merchants.
Choose how the value is provided
Next, select how Merchandising Cloud should obtain the value used by the filter.
Available Types include:
Type | How the value is determined |
|---|
Use product value | Uses the applicable value from the current product context. |
Provide value via API | Uses a value supplied through the implementation. |
Provide value manually | Lets you select a specific value available in your product data. |
When Provide value manually is selected, the values available in the next field depend on the Field you selected and the data configured in your store.
For example, selecting a colour-related product field allows you to choose from the colour values available in your catalogue.
Add additional filters if you need more than one rule. If no filtering is required, you can leave this section empty.
Save the Recommendation Engine
After configuring the Settings and any optional Filters, review the engine and select Save.
When creation is successful, Merchandising Cloud confirms that the Recommendation Engine has been created and opens Activation Settings. This final step provides the information required to place the recommendation on your storefront.
Activate the Recommendation Engine
The Activation Settings dialog appears after the Recommendation Engine is created. It provides the engine's Recommendation ID and the HTML snippet required to add the recommendation to the storefront.
Recommendation ID
The Recommendation ID uniquely identifies the Recommendation Engine. The interface also notes that this ID can be used to access the recommendation widget directly through the API.
Use the copy action beside the ID if you need it for your implementation.
Add the Recommendation Engine to the storefront
Under Code snippet to include on page, Merchandising Cloud generates an HTML snippet for the engine.
For example:
<div class="findify-element" id="product-findify-rec-3"></div>
Copy the generated snippet and paste it into the template of the page where you want the Recommendation Engine to appear.
You will see the instruction shown in the popup:
Copy and paste this HTML snippet into the template of the page you want it to appear on.
After copying the required activation details, select Close.
After you save a Recommendation Engine, it appears on the Recommendation Engines page. This confirms that the engine has been created successfully and gives you a central place to review the recommendations configured for the selected store.
Edit or manage a Recommendation Engine
After a Recommendation Engine is created, you can manage it directly from the Recommendation Engines page. Each engine has an Actions menu, accessed from the three-dot icon at the end of its row.
To manage an engine:
- Go to Recommendations → Recommendation Engines.
- Locate the Recommendation Engine you want to manage.
- Select the three-dot menu in the Actions column.
- Choose the required action.
From this menu, you can view the engine ID, preview the recommendation, edit its configuration, or delete the engine.
Option | What it does |
|---|
Show ID | Displays the Recommendation ID associated with the engine. Use this when you need to reference the engine in an implementation or API request. |
Preview | Opens a preview of the Recommendation Engine so you can review the products it returns before or while managing the configuration. |
Edit | Opens the Recommendation Engine configuration so you can update its settings. This includes the Page Type, Recommendation Type, product display settings, fallback configuration, and Filters. |
Delete | Removes the Recommendation Engine. Use this when the engine is no longer required. |
The State toggle on the Recommendation Engines page also lets you control whether an engine is active.