Marfeel Experiences: Experience Manager Overview

Experience Manager is where you create, configure, and deploy every experience. Regardless of format or blueprint, every experience passes through the same editor and the same four configuration tabs.

Create an experience

Blueprints are reusable patterns that provide sensible defaults for common use cases. Each blueprint pre-configures format, content source, and targeting based on a specific objective.

  1. Navigate to Experiences in the sidebar and click + New Experience.
  2. Select a blueprint from the creation gallery. Use the filters to narrow by type. See Blueprints to learn about creating and managing blueprints.
  3. The four-tab editor opens with sensible defaults from the selected blueprint.

Format tab

The Format tab defines how the experience appears on the page and when it activates.

What you configure here depends on the format you selected. Each format has its own placement model and interaction behavior. The format-specific documentation covers the details:

  • Inline: Selector-based placement within the page DOM (CSS or XPath).
  • Flowcards: Snap points, scroll-bound behavior, card icon.
  • Popups: Entry behavior (scroll-bounded or immediate).

Important: Once an experience is saved, its format cannot be changed. The format is set at creation time and becomes immutable. If you need a different format, create a new experience.

Triggers

Triggers control when the experience activates. Configure them under Advanced Settings in the Format tab. Not every trigger is available on every format. The available triggers depend on the format’s placement and interaction model.

Triggers are ideal for capturing user attention at the right moment — when the user is already engaged with content, has finished reading an article, or intends to leave the page.

Available trigger types:

Trigger What it does Common use cases
Scroll Absolute Activates after the user scrolls a specified number of pixels or percentage of the page. If both are set, activates when either condition is met Engagement-based prompts (subscription gates after 50% scroll). Note: Popups with this trigger use scroll-bounded entry (appear gradually as user scrolls)
Scroll Velocity Activates when the user scrolls a specified number of pixels within a time window. Negative pixel values detect upward scrolling Exit-intent patterns when users scroll too fast to be reading or scroll back up
Intersection Observer Activates when a specified element (identified by CSS selector) enters the viewport. Optional delay parameter End-of-article prompts when users reach the final paragraph, or hiding Flowcards when native ads appear
Timer Activates after a specified number of milliseconds since the Marfeel SDK loads Standard paywall implementations with timed content access
Click Activates when the user clicks a specific element (identified by CSS selector) Complex interactions like promoting an offer through a blocking banner, then moving it to a passive banner when user dismisses
Back Navigation Activates when the user hits the browser back button. Requires prior user interaction (click, tap, or keypress) Prevent bounce from Discover traffic by replacing article content with recommendations

AMP limitations: Percentage-based scroll and Intersection Observer triggers are not supported on AMP pages.

Flowcard-specific behavior: For Flowcards, each trigger must specify a target snap point. When the trigger fires, the Flowcard transitions to that snap point (Initial, Promoted, Active, Minimized, or Hidden). Intersection Observer triggers can be configured as recurrent (Every Time) for pages with infinite scroll, allowing the Flowcard to repeat behaviors as new content loads.

Without a trigger, the experience activates as soon as possible.

Content tab

The Content tab defines what the experience displays. What you see here depends on two things: the format you selected and the blueprint you started from.

For Recommender experiences, this is where you configure the ranking algorithm, time window, content restrictions, and CTR signals. You select which engine drives the recommendations (Most Read, Trending, Personalized, and others), set filters to narrow eligible articles, and configure fallback behavior.

For Page Transformations, the Content tab becomes the transformation editor where you select page elements and define class or HTML modifications.

For other blueprints, the Content tab provides format-specific configuration: HTML/Mustache layout editors, form field configuration, widget settings, or tag code snippets.

Visual formats also share a common set of style options in this tab:

  • Header and Icon: Title, subtitle, and icon for branding.
  • Colors: Primary, secondary, and background colors to match your site.
  • Hero Image: Header image from a specific URL or dynamically sourced.

Full details: Content and Recommender

Targeting tab

The Targeting tab defines who sees the experience. You build audience filters using two kinds of signals:

Editorial signals target based on the content the user is reading. Section, topic, tag, author, content type, IAB category. For example, you can show a live sports scores widget only to users reading articles in the “Sports” section, or display a subscription prompt exclusively on premium content.

User signals target based on who the user is and how they behave. Loyalty segment, geographic location, device, traffic source, custom variables, and conversion goals. For example, show a specific recommender only to users arriving from social networks, display an upgrade CTA only to loyal readers, or target a re-engagement prompt to users from a specific country.

You can combine both signal types. Filters within the same dimension combine with OR; different dimensions combine with AND. You can add the same dimension multiple times to create AND conditions (e.g., Topic includes “Sports” AND Topic does not include “Tennis”).

Goals

Target users based on conversion goals. Show a newsletter prompt only to users who haven’t subscribed. Show an upgrade CTA only to users who haven’t converted.

Full details: Targeting

Experiment groups

Built-in A/B testing. Split your audience into sticky variants and test different experiences against each other. Define experiment groups in Experiences Settings, then target each experience to a specific group.

Full details: A/B Testing and Experiment Groups

Delivery tab

The Delivery tab controls when, how often, and in what order the experience reaches your audience.

  • Scheduling: Set start date, end date, and custom dayweek schedules with hourly time blocks.
  • Frequency capping: Limit impressions per user per time window and set cooldown periods between consecutive impressions.
  • Volume and Distribution: For advertising and campaign-like experiences, you can set a total impression cap and control how it’s paced. For example, a branded
    content promotion with a budget of one million impressions: instead of burning through them in the first few hours, set the distribution to spread evenly over two weeks. The system handles the pacing automatically.

Full details: Delivery and Scheduling

Preview and test

Every experience has a live preview panel on the right side of the editor. Use it to verify how the experience looks and behaves before publishing.

How the preview works

The embedded preview doesn’t load your page directly. To display your site inside the editor, the page goes through a proxy that makes several modifications:

  • Browser security restrictions that prevent embedding pages from other domains are bypassed
  • Resource URLs (JavaScript files, stylesheets, images) are rewritten to load through the proxy
  • The Marfeel SDK is injected automatically to enable experience rendering

These modifications allow you to preview experiences without leaving the editor, but they also mean the preview is not an exact replica of your live page. Most of the time, this works seamlessly. However, certain site configurations — particularly around security policies and resource loading — can cause issues that only appear in preview mode.

If something works in live testing but fails in preview, or vice versa, the proxy modifications are often the reason. The most reliable test is always the live site using the Open in Browser feature.

Test URL

Enter a URL from your site in the preview bar to load that page as the test context. The experience renders on top of the page exactly as it would for a real user. Change the URL to test different pages, sections, or content types without leaving the editor.

Preview controls

Options menu

The three-dot menu to the right of the URL dropdown offers:

Option What it does
Mobile vs Desktop Switch between device simulations. Card format previews are limited to mobile.
JSON Display the raw JSON response. Useful for debugging Recommender API experiences.
Disable Triggers Bypass triggers so the experience loads immediately. Use this when focusing on visual adjustments.
Open in Browser Open a live debug preview in a new tab. Useful when the embedded preview has limitations (e.g., no ads rendered) or for sharing with colleagues.
HTML render endpoint Opens the fully rendered experience by itself in a new tab.
JSON endpoint Opens the raw JSON response in a new tab.
AMP endpoint Shows an example of how the experience should be implemented on AMP. The code is a guideline, not verbatim copy-paste.
RSS endpoint Opens the experience response in RSS format for API-driven recommenders.

Element selector

Element selection is available in two places: the selector field on the Format tab (for Inline experiences) and any time the Previewer enters Element Selection Mode. In element selection mode, you can click on any element in the preview to select it.

The selector field accepts both CSS selectors and XPath expressions. Any expression starting with / or .// is automatically recognized as XPath. Use CSS for straightforward class or ID targeting and XPath when you need to target elements by text content, DOM position, or parent-child relationships.

Once an element is selected:

  • The breadcrumbs bar shows the HTML hierarchy leading to that element
  • Click breadcrumbs to traverse up to parent elements or down to children, allowing you to select the right depth in the DOM
  • A dialog appears at the element’s bottom-right corner with options to:
  • Generate an optimized CSS Selector for Inline experiences (using Copilot)
  • Generate an HTML Layout for Recommender experiences (using Copilot)

Experience highlighting

When the experience loads successfully in preview, the viewport automatically scrolls to bring it into view and highlights it with a surrounding container for easy identification.

Live testing

To test an experience in a real browser or on your phone, click the Open in Browser button in the preview bar. This button automatically generates a test URL with all necessary parameters to render your experience, even if it’s unpublished or doesn’t match targeting rules.

The generated URL can be shared with colleagues for review or opened on your phone for mobile testing.

How it works: The system appends URL parameters to force the experience to render. The primary parameter is #mrfexp={EXPERIENCE_ID}, where the experience ID is taken from the editor URL (e.g., https://hub.marfeel.com/experiences/IL_1fLmaXLpQb6iQTbdVRa7HQ/edit/format). For preview-style behavior with auto-scroll and highlighting, the system uses #mrfdebug instead.

Note: These test parameters only work on web pages. They do not function on AMP due to platform limitations.

Test groups

When an experience uses experiment groups, the preview renders all variants simultaneously. You can see every version of the experience side by side without switching between groups.

Test multiple experiences simultaneously using comma-separated IDs:

https://www.yoursite.com/article#mrfexp=FC_fcS-HlkoJQfCMIPyvLqj-dg,IL_1fLmaXLpQb6iQTbdVRa7HQ

Only the specified experiences apply, even if other active experiences meet the targeting conditions.

Debug mode

Use #mrfdebug instead of #mrfexp to replicate preview behavior: the page automatically scrolls to and highlights the experience.

Important: The #mrfexp parameter works on Web only. It does not function on AMP due to platform limitations.

Save and publish

Experiences have three states: Draft, Published, and Archived.

Draft

Click Save as Draft to save your work without making the experience visible to users. The experience appears in the experience list with an Inactive status. You and your team can continue editing it, but it won’t render on your site.

Use drafts to prepare experiences in advance, get internal review, or iterate on configuration before going live.

Publish

Click Save & Publish to make the experience live. The targeted audience starts seeing it on your site. Changes go live within seconds — the CDN is purged on demand every time you publish.

This applies to both new experiences and edits to existing ones. If you modify a published experience and click Save & Publish, the updated version replaces the current one immediately.

Archive

Archive an experience to remove it from the main experience list. Archiving does not delete the experience — it hides it from the listing to reduce noise and keep focus on active work. You can restore an archived experience at any time.

Clone an experience

Clone an existing experience via the three-dot menu next to its name in the list, by selecting multiple experiences and using the Actions menu or from the drop-down action button when editing. The editor opens with all settings from the original and “(copy)” appended to the name. The clone is only created when you save.

Going deeper