Collections

Audience: Business users (marketers, lifecycle managers)

By the end of this article, you’ll understand:

What a collection is and how it enhances personalization
When and why to use collections in an AI Decisioning campaign
How to create and configure a collection in Hightouch
Where to use collections inside your actions

What is a collection?

A collection is a dynamic list of items — like products, content, or offers — that AID can use to personalize messages. Collections are pulled from catalogs you’ve already synced to Hightouch.

Instead of hardcoding one product into a message, a collection allows AID to select the best-fit item(s) for each user, based on behavior, preferences, or campaign performance.

When to use collections

Use collections when:

You want to test which items perform best in a message
You’re sending product recommendations (e.g., top sellers, “just for you”)
You want dynamic inserts like:
- “Recommended for you” product grids
- Personalized blog/article carousels
- Contextual discounts or promo bundles

You don’t need a collection if your message has fixed content that doesn’t change by user.

Navigating to Collections

To configure a collection:

Go to AI Decisioning > Collections
Click Add collection
Follow the three-step builder:
- Select a catalog
- Filter the catalog
- Finalize configuration

How to create a collection

Step 1: Select a catalog

Choose a data source that contains the items you want to recommend — for example, your synced Products or Content catalog.

Catalogs are created in Customer Studio. If you haven’t created a catalog yet, navigate to Customer Studio > Schema > Create > Catalog.

Step 2: Filter the catalog

Apply filters to limit which items the AI can select from. Common filters include:

In-stock only
Specific category or brand
Minimum discount threshold

If you skip this step, the AI will consider all items in the catalog.

Example: “In-stock spring accessories over $20”

Let’s say you want to create a collection of spring accessories that are currently available and cost more than $20. This could be used in a win-back email or seasonal campaign to re-engage lapsed shoppers with trending items.

You would set up your filters like this:

category is equal to Accessories
season is equal to Spring
inventory_status is equal to In Stock
price is greater than 20

This ensures the AI only recommends:

Spring-ready accessories
That are in stock
And priced above $20

Use this type of collection in a personalized message like: "We picked these trending spring accessories just for you."

Step 3: Finalize configuration

Name your collection.
- Use clear, descriptive names that reflect what the collection includes and how it’ll be used—like “In-stock Spring Accessories” or “Top Picks for Win-back.” This makes it easy to recognize and reuse collections across campaigns.
Configure callback behavior.
- Sometimes the ideal item might not be available at the time of send. You can choose:
  - Do not trigger the action if the item isn’t available
  - Send random items as a fallback

You can return to a collection later and adjust filters or fallback rules.

Eligibility (item-level targeting)

You can click into individual collections to add eligibility filters. Eligibility filters let you control which users are eligible to receive specific items from a collection. This ensures users only see recommendations that are relevant and appropriate for them.

You’ll define logic such as:

field is equal to static value
field is greater than/less than a defined value

Example:

Only recommend size-specific SKUs to users who’ve bought that size
Show luxury items only to loyalty-tier Platinum customers

Goal	Eligibility Filter Setup
Show only women’s products	`user.gender` is equal to `female`
Recommend premium items to high spenders	`user.lifetime_value` is greater than `500`
Target users who’ve viewed sunglasses	`user.last_viewed_category` is equal to `sunglasses`
Limit discounted items to bargain shoppers	`user.segment` is equal to `Value Seeker`
Personalize by location	`user.region` is equal to `Midwest`

You can stack multiple filters together using AND conditions, just like shown in the screenshot.

Use eligibility filters to avoid sending irrelevant product recommendations, especially in campaigns where precision and context matter (like loyalty, retention, or upsell flows).

Using collections in actions

Once your collection is created, you can apply it inside an action:

Go to AI Decisioning > Agents > Select an agent > Configuration > Select an Action
Navigate to the Advanced Configuration tab
Select a collection (e.g., “In-stock Products”)
Define how many items to recommend (e.g., 4 items per message)

Use a standard format for your product templates to make rendering easier across channels.

Preview and QA

Go to AI Decisioning > Agents > Select an agent > Configuration > Select an Action
Navigate to the Content tab to preview your content:
- Confirm that the correct number of items are inserted
- Test fallback logic
- Ensure templates display well with various item combinations