Bring your own catalog
Feed products from your PDP via feedProduct — once a product is in PRZIO, every recommendation activity can use it.
Recommendations that actually fit each shopper
Most viewed, similar items, who-viewed-this, AI theme expansion, or your own rules — fed by your real catalog and visitor behavior, served through one SDK call.
Platform
One activity, ten algorithms
Mix behavior signals, AI, and rules in a single recommendation activity. Same SDK call, different smarts behind it.
Behavior + AI + rules
AI-based, AI theme, most viewed, similar, who-viewed-this, recently viewed, last purchased, cart, or pure custom rules — pick one or mix several.
Include / exclude rules
Layer key/operator/value rules on top — match against the seed product, the visitor profile, or any catalog field with AND/OR logic.
Recommendation types
Pick the right algorithm for the placement
Each type has a clear job. Combine multiple in one activity to fill the carousel even when a visitor has limited history.
AI-based
Mixes most viewed, most purchased, cart abandon, categories, and visitor history — ranked by your configured AI provider.
AI theme expansion
The LLM infers a lifestyle theme from recently viewed categories and recommends related catalog categories.
Most viewed
Top products across the site or catalog, counted by unique sessions. Great for homepage and trending strips.
Similar product
Score on shared categories, tags, brand, and other identifying keys against the current product.
Who viewed this also viewed that
Co-view from sessions that viewed the seed product, with a category/tag fallback when traffic is sparse.
Recently viewed
Literal “your recently viewed” for the current visitor — perfect for cart, account, and post-search pages.
Based on recently viewed
Round-robin across categories the visitor browsed — diverse picks instead of more of the same item.
Based on last purchased
Score new candidates against categories/tags from the visitor's purchase history (cross-sell & replenishment).
Based on cart
Bring back cart abandoners with related products from the items they almost bought.
Custom criteria
Pure rules — match key/operator/value across categories, tags, brand, price, audience, or any catalog field.
Workflow
From catalog to live carousel
Three pillars — feed, configure, and embed. Each backed by the real PRZIO UI.
1. Feed
Push your catalog with one SDK call
Drop the snippet on each product page; PRZIO stores categories, tags, price, brand, and any extra keys you send.
- One SDK on every storefront page
- Catalog-scoped products with arbitrary keys
- View tracking via trackProductView
2. Configure
Pick algorithms and add safety rails
Choose one or many recommendation types, cap the count, exclude already-seen products, and decide what should happen when behavior data is missing.
- Combine algorithms — mixed results from each
- Include / exclude rules with AND/OR
- Match current product or visitor profile attributes
3. Embed
Render anywhere — one fetch, your UI
Call the recommendations endpoint with the activity ID; render with your own components, or use the PRZIO embed snippet for a drop-in carousel.
- Same activity works on web, app, or email
- Pass productId or userIdentifier to personalize
- JSON response — keep your design system intact
Use case
Recommendations in the PRZIO chatbot
The same project catalogs and recommendation activities can power your embedded assistant — help visitors discover products through conversation without leaving the chat.
Integration
Chatbot + product recommendations
Wire recommendation activities into conversational flows so the assistant can suggest relevant SKUs, explain choices, and deep-link to your storefront — all fed by the catalog you already send with feedProduct.
- One SDK stack for personalization, recommendations, and chat
- Consistent product metadata across carousel and assistant replies
- Ship faster by reusing activities instead of bespoke integrations
Integration
One snippet to feed, one call to render
The same SDK script powers personalization, popups, and product recommendations.
1. Load the SDK and feed products on each PDP
Replace YOUR_PROJECT_ID and YOUR_CATALOG_ID with values from the PRZIO catalog list.
<script src="https://app.przio.com/sdk.js"
data-project-id="YOUR_PROJECT_ID"></script>
<script>
if (window.PrzioSDK && window.PrzioSDK.feedProduct) {
PrzioSDK.feedProduct({
catalogId: 'YOUR_CATALOG_ID',
product: {
id: 'SKU-456',
title: 'Wireless Headphones',
category: 'Electronics:Audio:Headphones',
tags: ['sale', 'bestseller'],
price: 99.99,
brand: 'Acme',
thumbnails: ['https://mysite.com/images/sku-456.jpg']
}
});
}
</script>
2. Fetch recommendations for an activity
Use PrzioSDK.getRecommendations from the browser, or call the REST endpoint directly from your backend.
// Browser
const { recommendations } = await PrzioSDK.getRecommendations({
activityId: 'YOUR_ACTIVITY_ID',
productId: 'CURRENT_SKU', // optional — for similar / who-viewed-this
userIdentifier: 'visitor-id', // optional — for recently viewed / cart / purchase
limit: 10
});
// Backend
GET /api/sdk/recommendations
?projectId=YOUR_PROJECT_ID
&activityId=YOUR_ACTIVITY_ID
&limit=10
Ship your first recommendation today
Open the full guide for catalog setup, every recommendation type, include/exclude rules, and the complete API reference.
Open documentation