Boost recirculation with smart content recs.
A publisher’s guide to AI-driven content feeds — set up recommendations, track reader engagement, and deploy algorithms that boost recirculation and time on site.
Start Now →1The Wizard Process
From the Dashboard, select the "Content Recommendation" wizard. The 4 main steps are: Catalog → Tracking → Algorithms → Design & Targeting.
Step 1: Content Feed
Upload your content (XML/RSS/API) and map attributes (Title, Image, Date, Author). Tip: Use the "AI Auto-Interest" feature to scan your site for keywords if you lack tags.
Step 2: Interaction Tracking
Track engagement to build user profiles. Choose one method:
Option A: Site Containers (Visual)
Use the "Visual Selector" to click elements on your page (e.g., Author Name, Category Tag) to grab data without coding.
Option B: Programmatic JS (Developer)
<script>(self.personyze=self.personyze||[]).push(['Article Viewed', "ARTICLE_ID"]);</script><script>(self.personyze=self.personyze||[]).push(['Article Liked', "ARTICLE_ID"]);</script><script>(self.personyze=self.personyze||[]).push(['Article Commented', "ARTICLE_ID", 'Quantity', QUANTITY]);</script>
Step 3: Algorithm Selection
Choose the logic that drives the recommendations (e.g., "Most Popular", "Collaborative Filtering"). See the recipes below for examples.
2Advanced Filtering
Refine what content is shown using the "Final Touches" step.
Smart Logic Filters
- Exclude Read Content: "Exclude all Confirmed view" ensures users don't see articles they've already finished.
- Topic Match: "Only from Topic" -> "User read in past". Shows articles related to topics the user historically engages with.
- Author Match: "Only from Author" -> "Currently Reading". Shows more articles by the same writer.
Exclusion Rules
Strictly exclude content types, such as "Exclude all Commented" or "Exclude showing on current page" to avoid redundancy.
3Design, Placement & Targeting
Control where and how the recommendations appear.
Widget Design
Select a template (Grid, List, Slider) or customize the HTML/CSS. If you select JSON Feed, this step is skipped as you only configure the data output.
Placement
Use the Simulator to click a "Placeholder" on your live site. Alternatively, set the widget to appear as a Popup or slide-in. Note: You can add Multiple Widgets to a single page (e.g., "Trending" at top, "Personalized" at bottom).
Dynamic Badges
Add overlays like "Most Read", "New", or "Popular this Week" to increase CTR.
4Recipe 1: Homepage & Discovery
Engage readers immediately with relevant content.
1. Content Recommended for You
The gold standard. Displays articles matching the visitor's reading history and interests (e.g., "Politics" + "Europe").
2. Most Popular
Show what's trending across the site. Good for new visitors with no history.
3. Published Since Your Last Visit
A powerful retention tool. "Welcome back! Here are 5 new articles posted since you were here yesterday." Highly effective for news sites.
5Recipe 2: Article Page Recirculation
Keep readers on site after they finish an article.
1. Visitors Who Read This Also Read
Collaborative Filtering: "People who read this article also read X." Great for discovering related deep-dives.
2. Read Next (Contextual)
Recommend articles from the Current Category or with the same Tags. Keeps the user in their current flow/topic.
3. Managed Related Articles
Allows editors to manually curate specific recommendations for high-traffic articles.
6Recipe 3: High Engagement Boosters
Surface content that drives community interaction.
Most Commented
Show articles that are generating the most discussion. Encourages users to click and join the debate.
Most Liked / Favorited
Highlights "Crowd Favorites" based on explicit user feedback (Likes/Hearts).
7Recipe 4: Reader Retention
Personalize the experience to bring readers back.
Recently Viewed
A "History" widget allowing users to find articles they started reading but didn't finish.
Interest-Based Recommendations
If a user reads 3 articles about "Space", populate this widget ONLY with Space news, even if they land on the Sports page.
8A/B Testing Content Strategy
Optimize your circulation strategy.
Test Ideas
- Algorithm: Test "Collaborative Filtering" vs. "Content Similarity" (Tags).
- Time Frame: Test "Most Popular Last 24 Hours" vs. "Most Popular Last 7 Days".
- Design: Test a "Grid" layout vs. a "List" layout.
Personyze automatically tracks CTR and Time on Site to determine the winner.
9Performance & Analytics
Measure editorial success.
Key Metrics
- CTR (Click-Through Rate): The primary metric for content recommendations.
- Article Views: Total reads generated by the widget.
- Engagement Uplift: Compare "Time on Site" for users who clicked a recommendation vs. those who didn't.
AI Content Recommendations Wizard
See how Personyze enables you to inject personalized content widgets into any website using our Visual Editor. Watch the automated walkthrough below to see the process from template selection to live deployment.
Related Articles
Recommendations • Embedded
Select a template
Recommendation algorithm type
Most Popular
Trending content across site.
Recent
Newest content published.
For You
Personalized mix.
Most Commented
Highest engagement.
Trending Now
Hot in last 24h.
Rising Stars
Fastest growing views.
Similar by Tags
Based on user affinity.
Best Sellers
Top converting items.
Where to show content. You can choose popup or placeholder.
Content placements are defined using HTML selectors. By selecting a specific element on the page, Personyze can insert content at that exact location.
PERSONYZE QA SIMULATOR
Add action.
What should be executed when rules match.
Recommend interests
Recommend interests based on article views.
Content Recommendation Widget
Customizable responsive widget for personalized content.
Content Recommendations via JSON Feed
Client-side JSON structured data.
Playbook questions, answered.
Common questions about setting up content recommendations. Anything else, our team is one message away.
Do I need to tag all my articles manually?
No. Personyze can automatically extract tags, categories, and keywords from your page's meta tags or HTML structure. We also have an "AI Auto-Tagging" feature that analyzes the text content to generate interests dynamically.
How does it know what a user is interested in?
Every time a user views an article, Personyze records the associated tags (e.g., "Politics," "Technology") into their profile. Over time, we build an "Interest Graph" (e.g., User is 70% interested in Sports, 30% in Tech) to weight future recommendations.
Can I prioritize sponsored content?
Yes. You can use "Managed Recommendations" to pin sponsored posts to specific slots (e.g., Position 1) or boost articles with a specific tag (e.g., "Partner Content") to appear more frequently.
Does it work with video content?
Yes. As long as the video page has a unique URL and metadata (Title, Thumbnail), Personyze treats it just like an article. You can recommend "Videos You Might Like" based on viewing history.
Can I exclude old news?
Yes. You can add a filter to exclude content published more than X days ago (e.g., Date > Today - 30 days). This ensures recommendations remain fresh and relevant.
Does this work for anonymous visitors?
Yes. Personyze tracks session behavior immediately. If an anonymous user reads three articles about "Bitcoin," the system will instantly start recommending more crypto news within the same session.
Can I recommend content from different domains?
Yes. If you manage multiple sites (Cross-Domain), you can aggregate all content into a single feed and recommend articles from Site A to users on Site B to drive traffic across your network.
How do I prevent the same article from showing twice?
Enable the "Exclude Confirmed View" filter. Personyze tracks read history and will automatically remove articles the user has already visited to ensure they always discover something new.
Can I design the widget to look like my site?
Absolutely. You can use our visual CSS editor to match fonts, colors, and layout. Alternatively, you can use the JSON Output mode to send raw data to your own front-end template for pixel-perfect control.
What is "Collaborative Filtering" for content?
It's the "People who read X also read Y" logic. It doesn't look at the content topic itself, but rather at user patterns. It helps users discover content that is popular among people with similar reading habits.
Can I filter by Author?
Yes. You can set rules like "If user is reading an article by Author A, recommend more from Author A" or "Show latest posts from Author B" if the user follows them.
Does it impact page load speed?
No. The recommendation engine runs asynchronously. The rest of your page loads first, and the recommendation widget populates instantly afterwards, ensuring no delay in Core Web Vitals.
Can I A/B test different algorithms?
Yes. You can split traffic to test "Trending Now" vs "Personalized for You" to see which logic generates higher Click-Through Rates (CTR) and deeper engagement.
How often is the content feed updated?
Standard sync is daily, but we support real-time updates via RSS feeds or API pushes. This is crucial for news sites where breaking news needs to appear in recommendations immediately.
What metrics should I track?
For publishers, the key metrics are CTR (are people clicking?), Recirculation Rate (percentage of users who view another page), and Time on Site (did personalization increase session duration?).
Put your content recs to work.
You have the full playbook — now deploy AI-driven recommendations and watch recirculation and time-on-site climb.
Turn Visitor Intent into Revenue with
Smart Recommendations
A comprehensive platform that delivers tailored product and content recommendations at every stage of the customer journey—from first visit to loyal customer.
Step-by-Step Onboarding
Guided setup with a campaign wizard or a fully managed implementation by your dedicated success manager.
Customizable Widgets
Responsive templates that blend into your site and emails with a live visual editor and full CSS control.
API and Integrations
Works on any platform with server-side and client-side APIs, plus easy plug-and-play integrations.
Revenue & ROI Tracking
Prove your value. Advanced analytics track exactly how much revenue is generated by recommendations.
Set and Forget
Live learning. The engine continuously optimizes based on daily feedback to find the perfect match for every shopper.
Flexible Billing
A true pay-as-you-go model. Scale your personalization efforts without heavy upfront commitments.
Suggest the perfect product, every time
Stop guessing. Use Personyze’s intelligent recommendation engine to show the right products to the right shoppers. Boost AOV with "Frequently Bought Together" and AI-driven cross-selling strategies.
- Live Video Demo
- 30 Minutes
- Expert Q&A
