AI Product Recommendation

1. What is Magento 2 AI Product Recommendation?

Magento 2 AI Product Recommendation is an extension that helps merchants display relevant product suggestions across the storefront. It can use AI services, vector similarity, recommendation rules, trending products, and fallback products to generate recommendations.

2. Where can product recommendations be displayed?

Recommendations can be shown on supported storefront placements, including:

• Product pages
• Cart page
• Category pages
• Home page
• CMS blocks
• Checkout success page

Availability depends on the configured recommendation rules and layout placement.

3. Does the extension support both slider and grid layouts?

Yes. Recommendation rules can display products as either:

• Slider or carousel
• Responsive product grid

The number of visible products can be configured for desktop, tablet, and mobile devices.

4. Can I control which products are recommended?

Yes. Recommendation Rules allow you to control product selection by using:

• Store view
• Customer group
• Page placement
• Recommendation type
• Date range
• Product limit
• Product conditions
• Manual fallback products

5. What happens if the AI provider returns no recommendations?

The extension can use fallback strategies such as:

• Manual fallback products
• Rule-based fallback products
• Trending products
• Local vector search
• Generic catalog fallback

Fallback behavior depends on the configured connection mode and rule settings.

6. Does the extension support configurable products?

Yes. Configurable products are supported.

Where real product option selection is available, customers can select product attributes. If direct Add to Cart is not safe because required options must be selected, the card displays a Choose Options action that links to the product page.

7. Can customers add recommended products directly to the cart?

Yes. Simple and directly saleable products can be added to the cart from the recommendation card.

The extension uses Magento-compatible Add to Cart forms with:

• Product ID
• Form key
• Encoded return URL
• Magento cart initialization

8. Does the extension support Wishlist and Compare?

Yes. Recommendation cards can include Magento-compatible:

• Add to Wishlist
• Add to Compare

Guest wishlist behavior follows the standard Magento login flow.

9. Is the extension compatible with Hyvä Theme?

Yes. The extension includes separate frontend handling for Magento Luma and Hyvä Theme.

The Hyvä implementation avoids Luma-only jQuery dependencies and keeps product actions compatible with Hyvä storefront behavior.

10. Which AI connection methods are supported?

The extension supports multiple connection options, including:

• MageHQ AI Commerce API
• Direct provider API
• Local vector search
• Google Vertex AI integration
• Rule-based and manual fallback recommendations

The available options depend on your configuration and installed MageHQ AI modules.

11. Is Google Vertex AI required?

No. Google Vertex AI is optional.

You can use the extension with:

• MageHQ AI Commerce
• Direct provider credentials
• Local vector search
• Recommendation rules and fallback products

Google Vertex AI should only be enabled if you have a valid Google Cloud project and credentials.

12. Does Google Vertex AI generate additional costs?

It may. Google Cloud can charge for API requests, prediction services, event processing, catalog operations, and related cloud resources.

Merchants should review their Google Cloud billing and pricing before enabling Vertex AI.

13. What is Accuracy Filtering?

Accuracy Filtering allows you to set a minimum AI similarity score.

Products below the configured threshold can be excluded from AI-based recommendation results.

Fallback products can still be shown if fallback behavior is enabled.

14. What is Debug Mode?

Debug Mode helps administrators test and understand recommendation results.

When enabled, it can display information such as:

• AI Similarity
• Fallback source
• Popularity score
• Recommendation reason
• Minimum match threshold status

Debug information is hidden from normal storefront customers when Debug Mode is disabled.

15. What is the difference between AI Similarity and Popularity Score?

AI Similarity indicates how closely a product matches the current product or recommendation context.

Popularity Score is used for trending fallback products. It does not mean the product passed the AI similarity threshold.

16. Can I disable fallback recommendations?

Yes. You can configure whether fallback products should appear when no AI result meets the minimum match score.

If fallback is disabled and no suitable AI result is found, the recommendation block can return an empty state.

17. Does the extension include analytics?

Yes. The extension includes an AI Product Recommendation Dashboard that can display recommendation performance data such as:

• Impressions
• Clicks
• Click-through rate
• Recommendation Add to Cart events
• Recommendation source
• Placement performance
• Fallback usage
• Top recommended products
• Top recommendation rules
• Recent activity

Analytics availability depends on recorded recommendation-attributed events.

18. Does the dashboard count all Magento product views and cart events?

No. The dashboard is designed to count recommendation-attributed activity only.

Generic product views and unrelated cart events should not be included in recommendation performance metrics.

19. How are recommendation impressions tracked?

An impression is recorded after the recommendation block is successfully rendered.

The tracking system includes deduplication logic to reduce repeated impression events from reloads or repeated browser activity.

20. How are recommendation clicks tracked?

Clicks are tracked when customers click a recommended product image or product name.

The event can include context such as:

• Recommended product
• Source product
• Rule
• Placement
• Recommendation source
• Store

21. Does the extension support server-side tracking protection?

Yes. The tracking endpoint includes protections such as:

• POST validation
• CSRF protection
• Input validation
• Source normalization
• Server-side deduplication
• Rate limiting

22. Can I configure the number of recommended products?

Yes. The product count can be controlled through:

• Global Recommendation Count configuration
• Number of Products in a Recommendation Rule

If the rule product limit is greater than zero, it takes priority over the global default.

23. What happens when Number of Products is set to 0?

A value of 0 uses the global recommendation count instead of returning unlimited products or breaking the frontend layout.

24. Can I schedule recommendation rules?

Yes. Recommendation Rules can use:

• Date From
• Date To

This allows recommendations to run only during a selected campaign period.

25. Can I target specific customer groups?

Yes. A recommendation rule can be assigned to selected Magento customer groups, including guest customers.

26. Can I use different rules for product and cart pages?

Yes. Rules are matched by placement.

For example:

• A Product Page rule can show similar products.
• A Cart Page rule can show cart-based recommendations.
• A Home Page rule can show personalized or trending products.

27. Does the extension support multiple store views?

Yes. Recommendation rules and configuration settings can be assigned by store scope, depending on the Magento configuration field.

28. Are recommendation cards rendered by JavaScript?

Product cards are designed to use server-rendered HTML.

JavaScript is mainly used for:

• Lazy loading
• AJAX HTML injection
• Carousel behavior
• Magento frontend initialization

This helps preserve Magento product actions, form keys, and product rendering behavior.

29. Will recommendations slow down the product page?

The extension supports caching and asynchronous loading to reduce the performance impact.

Recommendation results can use:

• Cached product IDs
• Fast fallback products
• AJAX loading
• Provider fallback
• Configurable cache lifetime

Actual performance depends on the selected AI provider, store size, and cache configuration.

30. What happens if an AI provider is unavailable?

The extension can safely fall back to other recommendation sources if configured.

Examples include:

• Local vector search
• Manual fallback
• Rule fallback
• Trending products

The storefront should not display provider errors directly to customers.

31. Is the extension compatible with Magento production mode?

Yes. The extension is designed to support Magento production deployment, including:

• Dependency injection compilation
• Static content deployment
• Magento cache management

32. Does the extension require MageHQ AI Commerce modules?

MageHQ AI Commerce modules are required when using the MageHQ Subscription API connection mode.

Direct provider and local recommendation modes may use their own configuration paths.

33. Can API credentials be configured in the admin panel?

Yes. Direct provider and Vertex AI credentials can be configured in the Magento admin.

Sensitive credentials should only be accessible to authorized administrators.

34. Does the extension expose API keys on the frontend?

No. Provider credentials and service account information must remain server-side and are not included in frontend recommendation payloads.

35. Can I use the extension without AI?

Yes. The extension can still use Recommendation Rules, manual fallback products, and trending or catalog-based fallback logic when no AI provider is enabled.