## 💡 Prerequisites

**Knowledge** We recommend that you read the following guide:

  • [Widget Configurator](🔗): This will help you understand the process of creating a widget. We’ll be using a similar process to configure a Similar Products Widget.

**Technical Integration**

  • Make sure you have integrated [Recommendations and Pathways](🔗) with your site. Visit [this](🔗) article for more details.

# What is Similar Products Widget

This item-based widget enables you to recommend a range of similar products, and these recommendations are based on the product feed data.

Similar Products widget intelligently leverages product data to show the most contextually appropriate products. For example, if a user views a “_Simpsons backpack_,” the Similar Products widget might display recommendations for other backpacks with similar features.

Showing similar products encourages the shopper to explore your product catalog and discover other products they might be interested in buying. Similar Products Widgets can be placed on Product Description pages to create a **relevant browsing experience.**

Additionally, you can add Similar Products widgets to the Cart page to **upsell** customers while they check their cart items.

# Similar Products Widget Algorithm - How it Works

The Similar Products Widget is powered by the Similar Products algorithm. Here’s an overview of how this robust machine learning-based algorithm works:

  1. **Selecting Product Features** All the relevant product features are extracted from the product data you send us, and they are sent to the candidate generation algorithm. This feature data is used to train our algorithms for generating the best recommendation candidates.

  2. **Candidate Generation** We pass the collected product features information through the pre-trained [BERT](🔗) model and compute **product similarity scores**. The algorithm then predicts the potential top 200 similar products recommendations candidates.

This illustration summarizes how the Similar Products Algorithm ranks and builds the widget recall.

This illustration summarizes how the Similar Products Algorithm ranks and builds the widget recall.

# How to configure the Similar Products Widget

**Step 1**

Go to **`Pathways & Recommendations`** to view the Widget Configurator.

**Step 2**

Click **`Add Widget`**.

**Step 3**

Add the Widget Basic Info like **`Widget Name`** and **`Widget Description`**.

**Step 4**

Then, select the Primary Algorithm and Secondary Algorithm for the widget.

  1. **Primary Algorithm**: From the Primary Algorithm dropdown, select “**Similar Products**” as the Primary Algorithm.

  1. **Secondary Algorithm**: Choose the fallback algorithm for the Similar Products algorithm from the following options:

  • Best Seller

  • Trending Products

  • Recently Viewed

You can also set the **Result size** and **FL list** for the widget. When you finish adding the widget's basic information, click **Next Step** to proceed.

**Step 5**

Under the Product Grid tab, apply the merchandising rules for the Widget as described [here](🔗). Click **Save** to save the widget configurations.

**Step 6**

Once the widget is configured, note the _widget\_id_. The widget ID will be required in the [API call](🔗).

# How to access this Feature

  • **Already using Similar Products V1:** If you are already using Similar Products V1, we recommend that you upgrade to Similar Products V2. Please reach out to [Support](🔗) to access the latest feature version.

  • **Live Customers**: If you are already live, please reach out to [Support](🔗) to enable this feature.

  • **Integrating Customers**: Please reach out to your Technical Services representative who is helping with your integration.

## 📝Best Practices

  • Ensure that you have **integrated [Recommendations and Pathways pixels](🔗) properly**. These pixels capture customer behavioral data that help populate relevant widget recommendations. You can validate your pixel integration using the [Pixel Validator](🔗).

  • Ensure that you **send the right [product feed](🔗) information**. The Similar Product algorithm uses both user behavior and product information to build an intelligent model that generates similar product recommendations. Providing good data can improve the model's performance and lead to better product recommendations.

## Coming Up

  • Similar Products widget will support French and German language sites in the next feature phase.