Brand Concierge conversational experience

This guide provides a comprehensive implementation reference for AI-powered conversational experiences using Adobe Brand Concierge, integrated with Adobe Experience Platform (AEP) and Real-Time Customer Data Platform (RT-CDP). It is designed for solution architects, marketing technologists, and implementation engineers who need to deploy brand-safe conversational agents across digital properties.

It covers all viable approaches for deploying conversational experiences, from product advisory chatbots to full site navigation assistants, with guidance on when to choose each option. The plan addresses agent configuration, brand governance, content integration, deployment strategies, profile enrichment from conversation signals, and analytics optimization.

Brand Concierge enables brands to deploy intelligent conversational agents that understand brand voice, access approved product catalogs and content, deliver personalized recommendations based on real-time profile data, and capture intent and sentiment signals back into the unified customer profile. The result is a conversational experience that feels natural and on-brand while enriching the organization’s understanding of each customer.

Use case overview

Organizations increasingly seek to transform static digital experiences into dynamic, AI-powered conversations that guide customers through discovery, product selection, and purchase decisions. Adobe Brand Concierge addresses this by providing an orchestrated conversational AI layer that sits atop existing digital properties, powered by AEP Agent Orchestrator.

This pattern is distinct from traditional chatbot implementations because it is natively integrated with AEP’s unified profile, uses brand governance guardrails to ensure every response aligns with brand standards, and feeds conversational signals back into the customer data platform for downstream personalization and activation.

The target audience includes digital experience teams, e-commerce managers, content strategists, and marketing technologists who need to deploy intelligent conversational experiences that drive engagement, conversion, and profile enrichment.

Key business objectives

The following business objectives are supported by this use case pattern.

Deliver personalized customer experiences

Tailor content, offers, and messaging to individual preferences, behaviors, and lifecycle stage.

KPIs: Engagement, Conversion Rates, Customer Satisfaction (CSAT)

Learn more about delivering personalized customer experiences

Improve customer engagement

Increase interaction frequency and depth across all digital and physical touchpoints.

KPIs: Engagement, Time On (web) Page, Open Rates

Learn more about improving customer engagement

Increase conversion rates

Improve the percentage of visitors and prospects who complete desired actions such as purchases, sign-ups, or form submissions.

KPIs: Conversion Rates, Lead Conversion, Cost Per Lead

Learn more about increasing conversion rates

Acquire new customers

Expand the customer base through targeted acquisition campaigns, lookalike audiences, and paid media optimization.

KPIs: New Customers, Customer Acquisition Cost, Prospect/Lead Conversion

Learn more about acquiring new customers

Example tactical use cases

The following scenarios illustrate how this pattern can be applied in practice.

  • Product discovery assistant – Deploy a conversational agent on product listing pages that asks qualifying questions and narrows product recommendations based on customer needs, preferences, and budget
  • Guided comparison advisor – Help customers compare products side-by-side through natural dialogue, highlighting differences relevant to their stated priorities
  • Size and fit concierge – Guide apparel or footwear shoppers through size selection using conversational Q&A, reducing returns and increasing purchase confidence
  • Subscription or plan selector – Walk customers through service tier or subscription plan options with personalized recommendations based on usage patterns and stated needs
  • Site navigation assistant – Help visitors find relevant content, support resources, or product categories based on their stated intent, reducing bounce rates on complex sites
  • Pre-purchase consultation – Provide high-consideration purchase guidance (for example, electronics, financial products, insurance) through multi-turn conversations that build toward a recommendation
  • Loyalty program concierge – Help loyalty members discover rewards, understand tier benefits, and find redemption opportunities through conversational interaction
  • Re-engagement conversation – Initiate proactive conversational outreach to returning visitors based on previous browse history or abandoned cart items
  • Live agent escalation with context – Seamlessly hand off complex inquiries to live sales or support agents while preserving full conversation context and customer profile data
  • Post-purchase support and upsell – Engage customers after purchase with setup assistance, complementary product suggestions, and satisfaction check-ins through conversational channels

Key performance indicators

The following KPIs help measure the success of this use case pattern.

KPI
Description
Measurement approach
Conversation Engagement Rate
Percentage of visitors who initiate and sustain a conversation
Conversations started / eligible page views
Conversation Completion Rate
Percentage of conversations that reach a meaningful resolution
Completed conversations / conversations started
Conversational Conversion Rate
Percentage of conversations that lead to a desired action (purchase, sign-up, lead form)
Conversions from conversation / total conversations
Average Conversation Depth
Number of turns per conversation, indicating engagement quality
Average message count per session
Customer Satisfaction (CSAT)
Post-conversation satisfaction score from in-experience feedback
Survey responses or thumbs-up/down ratings
Recommendation Acceptance Rate
Percentage of product recommendations accepted or clicked
Recommendations acted on / recommendations served
Live Agent Handoff Rate
Percentage of conversations escalated to live agents
Handoffs / total conversations
Profile Enrichment Rate
Percentage of conversations that yield new intent or preference signals
Profiles enriched / total conversations
Revenue Influenced by Conversation
Revenue from purchases where a Brand Concierge conversation preceded the conversion
Attribution analysis on conversation-to-purchase journeys
Time to Resolution
Average duration from conversation start to resolution or handoff
Timestamp analysis across conversation events

Use case pattern

Brand Concierge conversational experience

Transform digital properties into AI-powered, brand-safe conversational experiences that guide customer discovery through natural dialogue, enrich profiles with intent and sentiment signals, and deliver personalized product recommendations.

Function chain: Agent Configuration > Brand Governance Setup > Content Integration > Conversational Experience Deployment > Profile Enrichment > Analytics & Optimization

Applications

The following applications are used to implement this use case pattern.

  • Brand Concierge – AI-powered conversational experience application providing the agent orchestrator, Product Advisor Agent, Site Advisory Agent, brand governance, and conversational analytics
  • Adobe Experience Platform (AEP) – Unified data foundation providing XDM schemas, identity resolution, real-time customer profiles, and data collection infrastructure for conversational signals
  • Real-Time CDP (RT-CDP) – Customer data platform providing real-time profile lookup for personalized conversations, audience segmentation from conversational signals, and profile enrichment with intent and sentiment data

Foundational functions

The following foundational capabilities must be in place for this use case pattern. For each function, the status indicates whether it is typically required, assumed to be pre-configured, or not applicable.

Foundational function
Status
What must be in place
Experience League reference
Administration & Governance
Required
Sandbox provisioned with Brand Concierge entitlement enabled; roles configured for conversational experience administrators, content managers, and analytics users; ABAC policies in place for conversational data containing PII or sensitive customer signals
Access control overview
Data Modeling & Preparation
Required
XDM schemas for conversational events (ExperienceEvent class with conversation-specific field groups capturing intent, sentiment, product interactions, and handoff events); profile schema extended with conversational preference and intent attributes; product catalog lookup schema for grounding recommendations
XDM System overview
Data Sources & Collection
Required
Web SDK or Mobile SDK configured with datastreams routing conversational event data to AEP datasets; Edge Network integration for real-time event capture during conversations; product catalog data ingested via source connectors or batch ingestion
Web SDK overview
Identity & Profile Configuration
Required
Identity namespaces configured for visitor identification (ECID for anonymous, CRM ID or email for authenticated); merge policy configured with edge activation for real-time profile lookup during conversations; identity linking rules for cross-device conversation continuity
Identity Service overview
Audience Definition & Segmentation
Assumed in Place
Audiences not required for core conversational deployment but needed for personalized conversation strategies (for example, high-value customer segments receive different conversation flows); streaming or edge evaluation recommended for real-time conversation personalization
Segmentation Service overview

Supporting functions

The following capabilities augment this use case pattern but are not required for core execution.

Supporting function
Status
Why it matters
Experience League reference
Computed / Derived Attribute Creation
Recommended
Aggregate conversational signals into profile-level attributes (for example, total conversations, dominant product interests, average sentiment score) for use in downstream segmentation and personalization
Computed attributes overview
Data Lifecycle Management
Recommended
Configure retention policies for conversational event data, manage consent for conversation recording and profiling, and support privacy deletion requests for conversation transcripts
Advanced Data Lifecycle Management overview
Data Usage Labeling & Enforcement
Recommended
Label conversational data fields containing PII, sentiment, or intent signals; enforce governance policies preventing sensitive conversational data from reaching unauthorized destinations
Data governance overview
Monitoring & Observability
Recommended
Monitor conversational event ingestion pipelines, track profile enrichment success rates, and alert on data flow failures that could affect conversation personalization quality
Observability Insights overview
Reporting & Analysis
Included
Analyze conversation performance, customer feedback, conversion attribution, and agent effectiveness using Brand Concierge built-in analytics and CJA for cross-channel conversation impact analysis
CJA overview

Application functions

This plan exercises the following functions from the Application Function Catalog. Functions are mapped to implementation phases rather than numbered steps.

Brand Concierge

Function
Implementation phase
Description
Agent Configuration
Phase 1: Agent Configuration
Configure the Brand Concierge agent orchestrator with agent specializations (Product Advisor, Site Advisory) and base behavior settings
Brand Governance Setup
Phase 2: Brand Governance Setup
Define brand voice, tone, messaging guardrails, approved content boundaries, and prohibited topics that shape all conversational interactions
Content Integration
Phase 3: Content Integration
Connect brand-approved content sources including AEM content, product catalogs, knowledge bases, and other trusted data for grounding responses
Product Advisor Configuration
Phase 3: Content Integration
Configure the Product Advisor Agent for personalized product recommendations, guided comparisons, and brand-aligned response delivery
Site Advisory Configuration
Phase 3: Content Integration
Configure the Site Advisory Agent to enhance navigation by adapting interactions based on visitor behavior and intent signals
Conversational Experience Deployment
Phase 4: Conversational Experience Deployment
Deploy conversational experiences across supported channels (web, mobile app, custom SDK) with text and voice interaction support
Low-Code Flow Management
Phase 4: Conversational Experience Deployment
Enable marketing teams to update conversational tone, flows, and content using low-code configuration tools
Conversational Profile Enrichment
Phase 5: Profile Enrichment
Enrich AEP customer profiles with intent, sentiment, product affinity, and behavioral signals captured during conversations
Conversational Analytics
Phase 6: Analytics & Optimization
Monitor engagement metrics, customer feedback, conversion data, and conversation quality via built-in analytics dashboards
Live Agent Handoff
Phase 4: Conversational Experience Deployment
Configure seamless handoff to live sales or support agents while preserving full conversation context

Real-Time CDP

Function
Implementation phase
Description
Real-Time Profile Lookup
Phase 4: Conversational Experience Deployment
Access real-time customer profile attributes and segment memberships to personalize conversational responses based on known customer data
Profile Enrichment
Phase 5: Profile Enrichment
Enrich profiles with computed attributes derived from conversational behavioral events (intent scores, sentiment trends, product affinity)
Audience Evaluation
Phase 5: Profile Enrichment
Evaluate audience membership based on conversational signals to enable downstream targeting of engaged conversational segments

Prerequisites

The following items must be in place before implementation begins.

  • [ ] Adobe Brand Concierge entitlement is active for the organization
  • [ ] AEP and RT-CDP licenses are provisioned with sufficient profile and event volume entitlements
  • [ ] Brand guidelines document available defining voice, tone, approved messaging, and prohibited topics
  • [ ] Product catalog or content repository prepared for integration (AEM assets, PIM data, or structured product feed)
  • [ ] Web properties identified for conversational experience deployment with technical access for SDK integration
  • [ ] Live agent infrastructure available if handoff is required (contact center platform, CRM integration)
  • [ ] Consent management framework in place for conversational data capture and profiling
  • [ ] Web SDK or Mobile SDK already deployed on target properties (or planned for concurrent deployment)
  • [ ] Stakeholder alignment on conversation scope (product advisory only, site navigation, or both)
  • [ ] Privacy and legal review completed for AI-powered conversational data capture and usage

Implementation options

The following sections describe different approaches for implementing this use case pattern.

Option A: Product Advisor deployment

Best for: E-commerce and retail organizations focused on guided product discovery, comparison, and recommendation experiences that drive conversion and average order value.

How it works:

The Product Advisor Agent is configured as the primary conversational specialization. It connects to the product catalog, understands product attributes and relationships, and guides customers through natural dialogue to arrive at personalized recommendations. The agent uses brand governance guardrails to ensure recommendations align with business priorities (for example, promoting in-stock items, highlighting margin-favorable products).

Product Advisor integrates with the real-time customer profile to access purchase history, browse behavior, and preference data, enabling recommendations that account for what the customer already owns, has previously considered, or is likely to need based on their profile. Conversations are captured as experience events and intent signals flow back into the profile for downstream use.

Key considerations:

  • Requires a well-structured product catalog with rich attribute data for effective recommendations
  • Product data must be kept current to avoid recommending out-of-stock or discontinued items
  • Brand governance must define how the agent handles competitive product mentions and price comparisons

Advantages:

  • Directly drives measurable revenue impact through guided purchase conversion
  • Reduces product return rates through better-informed purchase decisions
  • Captures high-value product affinity and intent signals for downstream personalization
  • Natural extension of existing e-commerce experiences

Limitations:

  • Requires ongoing product catalog maintenance and synchronization
  • Limited to product-centric conversations; site navigation questions may go unaddressed
  • Effectiveness depends on catalog data quality and completeness

Experience League:

Option B: Site Advisory deployment

Best for: Organizations with complex digital properties (media, financial services, healthcare, technology) where visitors need navigation assistance to find relevant content, resources, or self-service tools.

How it works:

The Site Advisory Agent is configured as the primary conversational specialization. It indexes the site content structure, understands page relationships and content categories, and adapts its guidance based on visitor behavior signals and stated intent. When a visitor engages, the agent interprets their needs and directs them to the most relevant content, tools, or resources.

Site Advisory uses real-time behavioral signals (current page, referral source, navigation path) combined with profile data (previous visits, content preferences, customer tier) to provide contextually relevant navigation assistance. This is particularly valuable on sites with deep content hierarchies, multiple product lines, or complex self-service workflows.

Key considerations:

  • Requires comprehensive content indexing and regular re-crawling as site content changes
  • Most effective on sites with significant content breadth where visitors commonly struggle to find what they need
  • Brand governance should define scope boundaries (which site areas the agent can navigate to)

Advantages:

  • Reduces bounce rates and improves content discoverability on complex sites
  • Captures navigation intent signals that reveal content gaps and user experience issues
  • Lower implementation complexity than Product Advisor (no product catalog integration required)
  • Provides analytics insights into what visitors are looking for but cannot find

Limitations:

  • Less directly tied to revenue conversion than product-focused conversations
  • Requires content to be well-structured and regularly updated for accurate guidance
  • May need frequent retraining as site structure evolves

Experience League:

Option C: Combined Product Advisor + Site Advisory deployment

Best for: Organizations that want a comprehensive conversational experience covering both product discovery and site navigation, typically large retail or B2C brands with extensive digital properties and diverse visitor intents.

How it works:

Both the Product Advisor Agent and Site Advisory Agent are configured within the Brand Concierge orchestrator. The agent orchestrator uses intent detection to route conversations to the appropriate specialization – product-related queries go to Product Advisor while navigation and content-finding queries go to Site Advisory. The orchestrator manages seamless transitions between specializations within a single conversation.

This approach provides the most complete conversational experience, handling the full range of visitor needs from “Help me find a product” to “Where can I check my order status?” Brand governance guardrails apply uniformly across both specializations, ensuring consistent brand voice regardless of conversation topic.

Key considerations:

  • Higher implementation complexity requiring both product catalog and content integration
  • Intent routing between specializations must be well-tuned to avoid misdirected conversations
  • Brand governance setup is more extensive to cover both product and navigation contexts

Advantages:

  • Provides the most comprehensive conversational experience for visitors
  • Single entry point handles diverse visitor intents without requiring separate interfaces
  • Cross-specialization conversations (for example, product question that leads to support navigation) handled naturally
  • Richest profile enrichment from diverse conversation signals

Limitations:

  • Highest implementation effort and ongoing maintenance
  • Requires coordination between product catalog and content teams
  • More complex testing and quality assurance requirements
  • Brand governance configuration is more involved

Experience League:

Option comparison

Criteria
Option A: Product Advisor
Option B: Site Advisory
Option C: Combined
Best for
E-commerce, product-driven conversion
Content-heavy sites, self-service navigation
Full-scope digital experience
Complexity
Medium
Low-Medium
High
Time to value
4-6 weeks
3-5 weeks
6-10 weeks
Revenue impact
High (direct conversion influence)
Medium (indirect via engagement)
Highest (both conversion and engagement)
Content requirements
Product catalog with rich attributes
Site content index
Both product catalog and content index
Profile enrichment
Product affinity, purchase intent
Navigation intent, content preferences
Full signal spectrum
Maintenance effort
Product catalog sync
Content re-indexing
Both ongoing

Choose the right option

Start by assessing your primary business objective and digital property characteristics:

  1. If your primary goal is driving product conversion and your digital property is commerce-focused, choose Option A (Product Advisor). This is the most common starting point for retail and e-commerce brands.

  2. If your primary goal is improving content discoverability and your site has deep content hierarchies or complex self-service workflows, choose Option B (Site Advisory). This is ideal for media, financial services, healthcare, and technology companies.

  3. If you need comprehensive coverage and have both product commerce and content navigation needs, choose Option C (Combined). Consider starting with one specialization and adding the second after the first is stable and optimized.

A phased approach is recommended for most organizations: deploy one specialization first, validate performance and gather learnings, then expand to the combined deployment.

Implementation phases

The following phases outline the recommended implementation sequence.

Phase 1: Agent configuration

Application function: Brand Concierge: Agent Configuration

Configure the core Brand Concierge agent orchestrator, including selecting agent specializations (Product Advisor, Site Advisory, or both), configuring base agent behavior, and establishing the connection between Brand Concierge and AEP for profile access and event capture.

Decision: Agent specialization selection

Determine which agent specializations should be activated for this deployment.

Option
When to choose
Considerations
Product Advisor only
Commerce-focused deployment targeting product discovery and conversion
Requires product catalog integration; fastest path to revenue impact
Site Advisory only
Content and navigation-focused deployment
Lower integration complexity; best for non-commerce sites
Both specializations
Comprehensive conversational coverage across product and content
Higher complexity; consider phased rollout starting with one

Decision: Conversation initiation model

Determine how conversations should begin on the digital property.

Option
When to choose
Considerations
Visitor-initiated (passive)
Default approach where the chat widget is available but does not proactively engage
Lower interruption risk; relies on visitor awareness of the chat option
Proactive engagement (triggered)
Agent initiates conversation based on behavioral signals (for example, extended dwell time, repeated page visits, cart hesitation)
Higher engagement rates but risks visitor annoyance if triggers are too aggressive; requires behavioral trigger tuning
Hybrid (passive with contextual prompts)
Chat widget is passive but displays contextual prompts based on page content or visitor behavior
Balanced approach; prompts guide without forcing engagement

Configure the agent

UI navigation: Experience Platform > AI Assistant > Brand Concierge > Agent Configuration

Key configuration details:

  • Define the agent name and description that will appear in the conversational interface
  • Select which AEP sandbox contains the customer profile and event data the agent should access
  • Configure the agent orchestrator to route queries between specializations based on intent detection
  • Set conversation session parameters (timeout duration, maximum conversation length, concurrent session limits)
  • Enable real-time profile lookup integration so the agent can access visitor profile data during conversations

Where options diverge:

For Option A (Product Advisor):
Enable the Product Advisor specialization and configure its connection to the product catalog data source. Set product recommendation parameters including maximum recommendations per response, product attribute display preferences, and comparison handling rules.

For Option B (Site Advisory):
Enable the Site Advisory specialization and configure its connection to the site content index. Set navigation parameters including content scope boundaries, page category handling, and deep-link generation preferences.

For Option C (Combined):
Enable both specializations and configure the orchestrator’s intent routing logic. Define routing rules that determine when a conversation should be handled by Product Advisor versus Site Advisory, and how transitions between specializations should be managed within a single conversation.

Experience League documentation:

Phase 2: Brand governance setup

Application function: Brand Concierge: Brand Governance Setup

Configure the brand governance guardrails that shape all conversational interactions. This includes brand voice and tone definitions, approved content boundaries, prohibited topics, response style guidelines, and escalation rules. Brand governance ensures every AI-generated response aligns with brand standards.

Decision: Governance strictness level

Determine how tightly the brand governance guardrails should constrain conversational responses.

Option
When to choose
Considerations
Strict governance
Highly regulated industries (financial services, healthcare, insurance) or premium brands requiring precise tone control
Limits conversational flexibility; may result in more frequent “I cannot help with that” responses; highest brand safety
Moderate governance
Most consumer brands where brand voice consistency matters but some conversational flexibility is acceptable
Good balance of brand safety and conversational naturalness; recommended starting point for most implementations
Flexible governance
Casual or lifestyle brands where conversational personality and engagement are prioritized
Most natural conversational feel; requires more ongoing monitoring for off-brand responses

Decision: Off-topic handling strategy

Determine how the agent should handle questions outside its configured scope.

Option
When to choose
Considerations
Redirect to scope
Agent acknowledges the question and redirects to topics it can help with
Maintains engagement but may frustrate visitors with legitimate off-topic needs
Handoff to live agent
Agent offers to connect the visitor with a human agent for off-topic questions
Best customer experience but requires live agent infrastructure and staffing
Graceful decline with resources
Agent explains it cannot help with that topic and provides links to relevant resources or support channels
Low-friction fallback; does not require live agent availability

Configure brand governance

UI navigation: Experience Platform > AI Assistant > Brand Concierge > Brand Governance

Key configuration details:

  • Define brand attributes: brand name, tagline, mission, values, and personality traits that inform conversational tone
  • Set tone parameters: formality level, humor tolerance, empathy level, and assertiveness for product recommendations
  • Configure approved content boundaries: topics the agent is authorized to discuss and topics that are explicitly prohibited
  • Define response format guidelines: maximum response length, use of lists versus prose, emoji policy, and link formatting
  • Set escalation triggers: conditions that should automatically route a conversation to a live agent (for example, complaint detection, repeated dissatisfaction signals, high-value customer identification)
  • Configure competitive mention handling: how the agent should respond when visitors ask about competitor products
  • Define disclaimer and legal notice requirements: mandatory disclosures for regulated industries

Experience League documentation:

Phase 3: Content integration

Application function: Brand Concierge: Content Integration, Product Advisor Configuration, Site Advisory Configuration

Configure the content sources that ground conversational responses in accurate, brand-approved information. This includes product catalog integration, AEM content connections, knowledge base imports, and content refresh schedules.

Decision: Product catalog integration method

Determine how product data should be provided to the Product Advisor Agent. (Option A and C only)

Option
When to choose
Considerations
AEP dataset integration
Product catalog is already ingested into AEP as a lookup dataset via source connectors
Leverages existing data infrastructure; keeps product data synchronized with profile data; requires foundational data modeling and collection to include product catalog
Direct feed integration
Product catalog exists in a PIM or commerce platform that can provide a structured feed
May offer more real-time inventory and pricing data; requires feed configuration and scheduling
AEM Content integration
Product content is managed in AEM and should serve as the authoritative product data source
Best for brands where AEM is the content hub; ensures consistency between web content and conversational responses

Decision: Content refresh frequency

Determine how often the agent’s content knowledge base should be updated.

Option
When to choose
Considerations
Real-time / near real-time
Product availability, pricing, or content changes frequently (for example, flash sales, inventory-sensitive retail)
Highest accuracy but higher infrastructure load; critical for inventory-sensitive recommendations
Daily refresh
Content changes are planned and scheduled (for example, editorial calendars, weekly promotions)
Good balance of accuracy and performance; suitable for most implementations
On-demand refresh
Content changes are infrequent and can be triggered manually when updates occur
Lowest overhead; suitable for static product catalogs or stable content sites

Configure content sources

UI navigation: Experience Platform > AI Assistant > Brand Concierge > Content Sources

Key configuration details:

  • Connect product catalog data sources with field mapping for product name, description, attributes, pricing, availability, images, and category hierarchy
  • Configure content indexing for site pages, knowledge base articles, FAQ content, and support documentation
  • Set content scope boundaries defining which content the agent can reference and which is excluded
  • Configure content fallback behavior when the agent cannot find relevant content to answer a question
  • Set up content quality rules: minimum content confidence threshold for inclusion in responses, citation requirements, and source attribution

Where options diverge:

For Option A (Product Advisor):
Focus on product catalog integration with rich product attribute mapping. Configure the Product Advisor Agent’s recommendation logic including how many products to suggest, how to handle out-of-stock items, how to present product comparisons, and how to incorporate customer profile data (purchase history, browse behavior) into recommendation ranking.

For Option B (Site Advisory):
Focus on site content indexing with page hierarchy mapping. Configure the Site Advisory Agent’s navigation logic including how to interpret visitor intent, which content categories to prioritize, how to handle ambiguous navigation requests, and how to adapt suggestions based on the visitor’s current page context and session behavior.

For Option C (Combined):
Configure both product catalog and site content sources. Ensure the content routing logic correctly assigns content to the appropriate specialization and that cross-references between product content and site navigation content are properly mapped.

Experience League documentation:

Phase 4: Conversational experience deployment

Application function: Brand Concierge: Conversational Experience Deployment, Low-Code Flow Management, Live Agent Handoff; RT-CDP: Real-Time Profile Lookup

Deploy the conversational experience on target digital properties, including channel configuration, widget customization, profile lookup integration for personalization, live agent handoff rules, and low-code tools for ongoing content management.

Decision: Deployment channel

Determine which channel(s) the conversational experience should be deployed to.

Option
When to choose
Considerations
Web (embedded widget)
Primary web property is the main customer touchpoint
Most common starting point; requires Web SDK integration; supports both anonymous and authenticated visitors
Mobile app (SDK integration)
Mobile app is a significant customer engagement channel
Requires Mobile SDK integration; consider screen real estate constraints for conversation UI
Custom SDK deployment
Conversational experience needs to be embedded in a custom application, kiosk, or non-standard digital property
Maximum flexibility; requires more development effort; suitable for in-store kiosks or proprietary platforms
Multi-channel deployment
Conversational experience needed across web, mobile, and other channels simultaneously
Highest reach; requires consistent brand governance across channels; conversation context should persist across channels when possible

Decision: Personalization depth for conversations

Determine how much customer profile data the agent should use to personalize conversations.

Option
When to choose
Considerations
Anonymous-only (session context)
Privacy-first approach or when most visitors are unidentified
Uses only in-session behavioral signals; no profile lookup required; suitable for anonymous product discovery
Profile-aware (authenticated visitors)
Visitors are typically logged in and personalized recommendations based on history add value
Requires real-time profile lookup via RT-CDP; significantly better recommendation quality for known customers
Progressive personalization
Blend of anonymous and authenticated with progressive profile building during conversation
Starts with session context; enriches as visitor provides information or authenticates; balances privacy and personalization

Decision: Live agent handoff configuration

Determine whether conversations should be escalable to live human agents.

Option
When to choose
Considerations
No handoff (self-service only)
AI agent can handle all expected conversation types, or live agents are not available
Simplest deployment; may frustrate visitors with complex needs; suitable for low-risk, product-browsing scenarios
Rule-based handoff
Specific triggers should escalate to live agents (for example, complaint detection, high-value customer, complex inquiry)
Predictable escalation behavior; requires defining escalation rules and triggers; needs live agent platform integration
Visitor-requested handoff
Visitors can request a live agent at any point in the conversation
Best customer experience; requires always-available agent staffing or queue management; conversation context must transfer

Deploy the conversational experience

UI navigation: Experience Platform > AI Assistant > Brand Concierge > Deployment

Key configuration details:

  • Configure the conversational widget appearance: position, color scheme, avatar, welcome message, and interaction style (text, voice, or both)
  • Integrate with Web SDK or Mobile SDK for event capture and profile resolution
  • Configure real-time profile lookup to access customer attributes, segment memberships, and recent activity during conversations
  • Set up live agent handoff integration with the contact center platform, including context transfer protocol, queue routing, and agent notification
  • Enable low-code flow management tools for marketing teams to update conversation starters, promotional messaging, seasonal content, and flow variations without developer involvement
  • Configure conversation session persistence rules: how long conversation history is retained, whether conversations can resume across sessions, and cross-device conversation continuity

Experience League documentation:

Phase 5: Profile enrichment

Application function: Brand Concierge: Conversational Profile Enrichment; RT-CDP: Profile Enrichment, Audience Evaluation

Configure the capture and enrichment pipeline that feeds conversational signals back into the AEP unified customer profile. This includes mapping conversation events to XDM, extracting intent and sentiment signals, creating computed attributes from conversational data, and building audiences based on conversational behaviors.

Decision: Conversational signal capture scope

Determine which conversational signals should be captured and written to the customer profile.

Option
When to choose
Considerations
Core engagement signals only
Minimal profile enrichment; capture conversation start, end, duration, and completion status
Lowest data volume; sufficient for basic analytics; limited personalization value
Intent and preference signals
Capture inferred product interests, stated preferences, and topic categories discussed
High personalization value; moderate data volume; most commonly recommended
Full signal capture
Capture intent, sentiment, product interactions, recommendation responses, handoff events, and feedback scores
Richest profile enrichment; highest data volume; enables advanced analytics and ML-driven personalization

Decision: Audience creation from conversational data

Determine whether audiences should be created based on conversational behaviors for downstream activation.

Option
When to choose
Considerations
No conversational audiences
Conversational data used for analytics only, not for audience activation
Simplest approach; suitable if conversations are supplementary to existing engagement channels
Intent-based audiences
Create audiences based on stated product interests or navigation intents from conversations
Enables retargeting visitors who expressed interest but did not convert; high-value for commerce
Behavioral audiences
Create audiences based on conversation engagement patterns (for example, high engagement, abandoned conversation, repeated visits)
Enables conversation-informed journey orchestration and cross-channel follow-up

Configure profile enrichment

UI navigation: Experience Platform > Customer > Profiles > Computed attributes (for derived signals); Customer > Audiences > Create audience (for conversational audiences)

Key configuration details:

  • Map conversational events to XDM ExperienceEvent schema fields capturing conversation ID, message count, topics discussed, products referenced, sentiment scores, and resolution status
  • Configure Brand Concierge profile enrichment to write intent and preference signals to the unified profile
  • Create computed attributes from conversational event data: total conversations (lifetime), dominant product category interest (30 days), average sentiment score (90 days), conversation-to-purchase conversion rate
  • Define streaming or batch audience segments based on conversational signals for downstream activation (for example, “Visitors who discussed Product Category X in the last 7 days but did not purchase”)
  • Validate profile enrichment by looking up sample profiles to confirm conversational attributes are populated

Experience League documentation:

Phase 6: Analytics and optimization

Application function: Brand Concierge: Conversational Analytics

Set up analytics dashboards and reporting for measuring conversational experience performance, identifying optimization opportunities, and tracking KPIs. This includes Brand Concierge built-in analytics, optional CJA integration for cross-channel conversation impact analysis, and ongoing optimization workflows.

Decision: Analytics depth

Determine what level of conversational analytics is needed.

Option
When to choose
Considerations
Built-in Brand Concierge analytics
Standard reporting on conversation volume, engagement, satisfaction, and conversion is sufficient
Fastest to activate; covers core KPIs; limited cross-channel correlation
Brand Concierge + CJA integration
Cross-channel analysis needed to understand how conversations influence broader customer journeys
Requires CJA connection and data view setup; enables attribution analysis across conversation and other channels
Full analytics stack (Brand Concierge + CJA + custom dashboards)
Executive-level reporting, advanced attribution modeling, and custom audience creation from analytics insights
Highest analytical capability; requires CJA expertise; enables data-driven conversation optimization

Configure analytics and optimization

UI navigation: Experience Platform > AI Assistant > Brand Concierge > Analytics; Analytics Platform > Workspace (for CJA)

Key configuration details:

  • Review Brand Concierge built-in analytics dashboards: conversation volume trends, engagement rate, completion rate, CSAT scores, recommendation acceptance rate, and handoff frequency
  • Configure CJA connection to include conversational event datasets for cross-channel analysis (if choosing CJA integration)
  • Build CJA workspace analysis for conversation-to-conversion attribution, identifying which conversation topics correlate with purchase behavior
  • Set up conversation quality monitoring: track topics where the agent struggles, common unanswered questions, and sentiment trends over time
  • Define optimization workflows: regular review cadence for brand governance updates, content refresh triggers, and conversation flow improvements based on analytics insights

Experience League documentation:

Implementation considerations

The following sections cover guardrails, common pitfalls, best practices, and trade-off decisions to keep in mind during implementation.

Guardrails and limits

  • Brand Concierge conversational experiences are subject to AI response generation rate limits; concurrent conversation capacity depends on entitlement tier
  • Real-time profile lookup during conversations is subject to Profile API rate limits per sandbox – Real-Time Customer Profile guardrails
  • Conversational event data ingestion follows standard AEP streaming ingestion limits – Ingestion guardrails
  • Product catalog size and content index volume are subject to Brand Concierge content integration limits
  • Maximum of 25 computed attributes per sandbox applies to conversational signal aggregations – Computed attributes guardrails
  • Maximum of 4,000 segment definitions per sandbox applies to conversational audiences – Segmentation guardrails

Common pitfalls

  • Insufficient brand governance definition: Deploying without thorough brand governance configuration results in off-brand responses that damage customer trust. Invest significant time in Phase 2 defining tone, boundaries, and escalation rules before deployment.
  • Stale product catalog data: Product Advisor recommendations based on outdated inventory, pricing, or availability data frustrate customers and erode confidence. Establish automated content refresh pipelines with validation checks.
  • Overaggressive proactive engagement triggers: Setting behavioral triggers too aggressively (for example, triggering conversation after 3 seconds on page) annoys visitors and increases bounce rates. Start with conservative triggers and tune based on engagement data.
  • Neglecting anonymous visitor experience: Focusing personalization only on authenticated visitors ignores the majority of traffic. Design conversation flows that provide value to anonymous visitors using in-session behavioral signals.
  • Skipping profile enrichment configuration: Deploying conversations without capturing signals back to the profile wastes valuable intent and preference data. Configure profile enrichment in parallel with deployment, not as an afterthought.
  • Ignoring live agent handoff experience: Poor handoff experiences (lost context, repeated questions, long wait times) damage the overall conversational experience more than not offering handoff at all. Test the full handoff flow end-to-end before launch.

Best practices

  • Start with a single agent specialization (Product Advisor or Site Advisory) and expand after establishing baseline performance.
  • Conduct brand governance workshops with marketing, legal, and customer experience stakeholders before configuring guardrails.
  • Use progressive personalization: start conversations with session-context-based responses and deepen personalization as the visitor provides information or authenticates.
  • Implement A/B testing of conversation starters, prompts, and recommendation presentation formats using the low-code flow management tools.
  • Schedule regular (weekly or biweekly) review of conversation analytics to identify content gaps, common failure points, and optimization opportunities.
  • Create a feedback loop between conversational analytics and brand governance updates – use conversation data to refine tone, add new approved topics, and adjust escalation rules.
  • Monitor conversation sentiment trends as an early warning system for product issues, site problems, or brand perception shifts.
  • Design conversation flows that naturally capture profile-enriching signals without making the interaction feel like an interrogation.

Trade-off decisions

NOTE
The following trade-off decisions should be evaluated based on your organization’s specific requirements and constraints.

Conversation personalization depth vs. privacy simplicity

Deeper profile integration enables more personalized and effective conversations, but increases data collection complexity, consent requirements, and privacy compliance burden.

  • Deep personalization favors: Higher conversion rates, better recommendation quality, richer profile enrichment, and more engaging conversations for returning customers
  • Privacy simplicity favors: Faster deployment, simpler consent management, lower regulatory risk, and a privacy-first brand positioning
  • Recommendation: Start with progressive personalization that works well for anonymous visitors and adds profile-based personalization for authenticated sessions. This provides value at all identification levels while keeping privacy compliance manageable. Implement consent capture for conversational profiling aligned with existing consent frameworks.

Brand governance strictness vs. conversational naturalness

Strict brand governance guardrails ensure every response aligns with brand standards, but overly rigid constraints make conversations feel robotic and reduce engagement.

  • Strict governance favors: Brand safety, regulatory compliance, consistent messaging, and predictable agent behavior
  • Flexible governance favors: Natural conversation flow, higher engagement, better customer satisfaction, and the ability to handle a wider range of queries
  • Recommendation: Begin with moderate governance and tighten or loosen based on conversation analytics. Monitor the rate of “I cannot help with that” responses as an indicator of over-restriction. Use the low-code flow management tools to iterate on governance settings quickly without developer involvement.

Real-time content refresh vs. system performance

Real-time content synchronization ensures the agent always has current product and content data, but continuous refresh consumes more infrastructure resources and can introduce latency.

  • Real-time refresh favors: Accuracy for inventory-sensitive recommendations, time-sensitive promotions, and rapidly changing content
  • Scheduled refresh favors: System stability, predictable resource consumption, and lower infrastructure costs
  • Recommendation: Use daily content refresh as the default, with near-real-time refresh only for inventory availability and pricing data that materially affects the customer experience. Monitor content accuracy metrics to determine if the refresh frequency is adequate.

Comprehensive signal capture vs. data management overhead

Capturing every conversational signal provides the richest profile enrichment and analytics, but increases data volume, storage costs, and governance complexity.

  • Full signal capture favors: Advanced analytics, ML model training, comprehensive profile enrichment, and maximum downstream personalization value
  • Selective capture favors: Lower storage costs, simpler data governance, faster profile lookup performance, and easier compliance with data minimization principles
  • Recommendation: Start with intent and preference signal capture (the middle ground) and expand to full signal capture only after validating that the additional data creates measurable downstream value. Apply dataset expiration policies to conversational event data to manage storage growth.

The following resources provide additional information for implementing this use case pattern.

Brand Concierge

Adobe Experience Platform

Data collection and integration

Identity and profile

Audiences and segmentation

Data governance and privacy

Monitoring and observability

Analytics and reporting

Guardrails

recommendation-more-help
045b7d44-713c-4708-a7a6-5dea7cc2546b