Contextual Embeds: How They Work and How to Configure Them

Overview:

This guide is for product teams, engineers, and growth teams implementing Genuin video experiences across websites and mobile apps who need feeds to adapt dynamically based on user and page context.

Use Contextual Embeds when you want to:

Show different videos on the same embed depending on page, search, or user intent
Deliver highly relevant video feeds for anonymous and logged-in users
Adapt content in real time without changing Placement configuration
Personalize experiences across web and app using runtime signals

Modern users expect content to respond to what they are doing right now, not just who they are. Contextual Embeds allow brands to deliver intent-aware, adaptive video experiences, while keeping governance, structure, and content rules centralized in the Brand Control Center.

First, Clarifying the Difference: Placements vs Embeds

At a high level:

Embeds are rendering containers - how Genuin content is visually embedded on a page.
Placements are experience definitions - why, where, and under what rules that content appears.

Think of it this way:

Embeds define the player. Placements define the strategy.

What Is a Placement?

A Placement is a context-aware content experience designed for a specific business moment, journey stage, or environment.

It focuses on:

Intent
Targeting
Personalization
Governance
Activation logic

What Placements control

Placements answer questions like:

Where is this shown in the customer journey?
What content is allowed here?
How is it personalized?
How does it align with brand, category, or objective?

What Is an Embed?

An Embed is a technical display unit used to surface Genuin content on a page or surface.

It focuses on:

Visual structure
Player behavior
Interaction controls

What Embeds control

Embeds answer questions like:

What feed is shown?
What does it look like?
What UI elements are visible?
How does the video behave inside the frame?

What Are Contextual Feeds?

A Contextual Feed is a dynamically generated video stream powered by Genuin Adaptive Intelligence, using real-time signals from a user’s environment and behavior.

These signals may include:

Page context (what content or category the user is viewing)
Search keywords entered on your site or app
User attributes such as interests or profile data
Geographic signals like latitude and longitude

Genuin processes these inputs together to determine intent and relevance, then surfaces videos most likely to resonate in that specific moment.

The result: a feed that adapts continuously as user context changes.

How Contextual Feeds Work

Contextual Feeds operate on a multi-signal recommendation model, where different types of context are evaluated independently and combined intelligently.

At any given time, the system prioritizes the strongest available signal, ensuring high-quality recommendations even when some data points are missing (for example, anonymous or first-time users).

Supported Context Inputs

Page Context

The page or screen a user is currently viewing, such as:

“SUV Listings”
“Car Loan Guides”
“Winter Fashion Collection”

This helps align video content with the surrounding editorial or commerce experience.

Search Query Context

Keywords or phrases entered by the user within your site or app.These signals indicate explicit intent and are given high relevance in feed generation.

User Profile Context

Includes available user attributes such as:

Age range
Interests
Bio or stated preferences

When present, these signals help personalize feeds across sessions.

Geographic Context

Latitude and longitude, typically derived from IP or device data, are used to:

Localize recommendations
Surface regionally relevant or nearby content

Video Selection and Ranking Logic

Contextual Feeds use a multi-stage selection and ranking process to balance relevance, freshness, and engagement.

1. User Profile–Based Selection

Videos are first matched against known user interests and attributes.
If no strong matches exist (cold-start scenarios), the system automatically falls back to contextual or location-based signals.

Engagement weighting within this stage:

Shares (highest influence)
Sparks/Reactions (medium)
Views (baseline)

2. Context-Based Selection

Text-based inputs (page context, search terms, bios) are analyzed to extract:
- Keywords
- Topics
- Implied intent
Location signals are applied where relevant.
These signals are converted into structured queries and executed via OpenSearch to retrieve matching videos.

3. Final Ranking and Feed Output

Each candidate video is assigned a final score using a balanced weighting model:

index.html

Final_score = 0.34 × Context_score
              + 0.33 × Recency_score
              + 0.33 × Popularity_score

Videos are then sorted in descending order of this score and delivered as a ranked feed.

This ensures users see content that is:

Contextually relevant
Recently published
Proven to engage other users

Contextual Feed Configuration Parameters

To enable contextual recommendations, your embed or SDK implementation passes a structured payload including:

Brand and community scope
Optional user identifiers and interests
Contextual feed flag
Ranking weights
Page and location context

These parameters allow fine-grained control over how feeds behave, without hardcoding content decisions.

index.html

{ 
  "brand_id": int, 
  "communities": [ 
   { 
     "community_uuid": "string", 
     "groups": ["string", "string", "string"] 
   }, 
   { 
     "community_uuid": "string", 
     "groups": ["string", "string", "string"] 
   } 
  ], 
  "user_uuid": "string", 
  "user_interest": ["string", "string", "string"], 
  "page_session": "string", 
  "genuin_user": true, 
  "limit": int, 
  "contextual_feed": true, 
  "user_weight": float, 
  "recency_weight": float,   
  "popularity_weight": float, 
  "page_context": { 
    "page_context": "string", 
    "compare_with": "video/community/brand/group/location",
    "location": [lat, long], 
    "location_radius": int 
  } 
}

How the compare_with Parameter Works

The compare_with field determines how contextual similarity is evaluated:

Value	Behaviour
community	Matches videos from communities with similar descriptions
brand	Surfaces videos linked to related brands
video	Finds videos with similar contextual signals
location	Filters content within a defined geographic radius

If contextual_feed is set to false, Genuin defaults to standard (non-contextual) feed logic.

Contextual Embed Activation Flow

Once enabled, the system follows this sequence:

Contextual feed flag is validated
Videos are filtered by brand_id and scope
A context vector is generated from available inputs
Videos receive contextual relevance scores
Final ranking is applied
A personalized feed is returned to the embed

This flow allows every embed to respond dynamically to user behavior without manual intervention.

How to Configure Contextual Feeds Across Platforms

Contextual Feeds can be implemented using Genuin’s SDKs across platforms:

Each SDK provides:

Step-by-step integration guidance
Configuration examples
Platform-specific best practices

Refer to the respective SDK documentation for implementation details.

Summary: Why Contextual Embeds Matter

Contextual Embeds enable media and commerce brands to move beyond static content blocks and deliver real-time, intent-aware video experiences.

By combining page signals, user behavior, and location data - within a governed, configurable framework - Contextual Feeds help you:

Increase relevance and engagement
Improve content discovery
Personalize experiences without sacrificing control
Scale intelligent video across all owned digital surfaces

This ensures every embed feels purposeful, timely, and aligned with the user’s journey - while remaining fully brand-safe and auditable.

Get Started

User Guides

SDK References