AD0-E454 — Adobe Audience Manager Architect

DIL (Data Integration Library) Implementation

• DIL is the primary JavaScript API for collecting data from web pages and sending it to AAM data collection servers
• DIL can collect page-level data, query string parameters, referrer URLs, and custom key-value pairs
• DIL version 8.0+ integrates natively with Adobe Experience Platform Identity Service for cross-solution identity
• Server-side forwarding (SSF) is preferred over DIL when Adobe Analytics is present, reducing duplicate calls
• DIL GA modules extend functionality for specific use cases like Google Analytics data ingestion

Server-Side Forwarding (SSF)

• SSF forwards Analytics data directly from Adobe data collection servers to AAM, eliminating client-side DIL calls
• Must be enabled both in Analytics Report Suite settings AND in AAM via the Admin UI
• SSF requires Adobe Experience Cloud Identity Service (ECID) to be implemented
• SSF sends Analytics hits in real time, enabling trait qualification on every page view
• Pixel calls are not needed when SSF is active — all Analytics variables are forwarded automatically

Inbound Batch Data Ingestion

• Batch data is ingested via Amazon S3 or FTP using a defined file naming convention
• File format: .sync or .overwrite; .sync adds/updates records, .overwrite replaces all data for a user
• Inbound files must follow the format: <dpid>_<dpid_target>_<timestamp>.<format>
• AAM supports tab-delimited flat files and GZIP-compressed files for batch ingestion
• Onboarding Status Reports provide diagnostics on file processing success, errors, and trait qualification counts
• Customer data records are matched to AAM UUIDs via declared IDs or device IDs

Real-Time Data Collection and Pixel Calls

• Event calls (pixel calls) send key-value pairs to AAM for real-time trait qualification
• Pixel calls use the format: https://pixel.everesttech.net/... or the AAM DCS endpoint
• DCS (Data Collection Server) is the real-time component that processes trait qualification and returns segment information
• AAM supports GET and POST calls to the DCS for server-to-server integrations
• ID synchronization pixels align partner IDs with AAM UUIDs for people-based targeting

Mobile SDK and App Data Collection

• Adobe Mobile SDK supports AAM data collection for iOS and Android applications
• Mobile SDK sends lifecycle data, custom context data, and analytics hits that can be forwarded to AAM
• Audience Manager module within the SDK handles trait qualification and segment return
• Mobile implementations can pass declared IDs to bridge authenticated and anonymous profiles
• Profile merge rules must account for device-based vs. cross-device authenticated profiles in mobile contexts

Trait Types and Configuration

• Rule-based traits qualify users in real time based on key-value pairs sent to the DCS
• Onboarded traits qualify users from batch-ingested offline data files
• Algorithmic traits are generated by Look-Alike Modeling and cannot be manually edited
• Folder traits automatically roll up all rule-based and onboarded traits within a folder hierarchy
• Third-party traits are sourced from the Audience Marketplace and cannot be edited by the buyer
• Trait recency (R) and frequency (F) can be applied to rule-based traits using operators like >, <, ==, and between

Trait Qualification and TTL

• Trait Time-to-Live (TTL) determines how long a user remains qualified for a trait after the qualifying event
• Default trait TTL is 120 days; can be set from 1 to 180 days for rule-based traits
• Onboarded traits use a default TTL of 180 days
• Trait recency is measured in days; a trait with recency of 7 means the user qualified within the last 7 days
• Trait frequency sets the minimum number of times a signal must be seen for trait qualification
• Users are removed from traits when TTL expires unless they re-qualify

Segment Building and Logic

• Segments use Boolean logic: AND, OR, AND NOT operators to combine traits
• Segment recency and frequency can be applied at the segment container level independently of trait-level R/F
• Segments can include up to 100 traits directly; nested segments count as single traits
• Segment TTL is controlled by the longest trait TTL within the segment
• Sequential segmentation can be approximated using recency/frequency rules across multiple trait containers
• Segment population reports show real-time (1-day) and total (30-day) addressable audience sizes

Algorithmic Modeling and Look-Alike

• Look-Alike Models find users who resemble a baseline segment by analyzing shared trait patterns
• The baseline segment (seed audience) should have at least 1,000 unique users for reliable model accuracy
• Model accuracy and reach are inversely related — higher accuracy means a smaller, more precise audience
• Algorithmic traits are created from model output at defined reach/accuracy thresholds
• Models refresh every 8 days automatically; manual refresh is also available
• TraitWeight is the score (0-100) assigned to each trait indicating its contribution to model differentiation

Profile Merge Rules (PMR)

• PMRs define which device graphs or profile options AAM uses when evaluating segment membership
• Options include: Current Authenticated Profiles, Last Authenticated Profiles, Device Profile
• Co-op Device Graph (Adobe Device Co-op, now retired) and external graphs (LiveRamp) were supported options
• PMRs affect addressable audience calculations — authenticated rules produce smaller but higher-quality audiences
• A segment can only use one PMR; choosing the right rule is critical for cross-device targeting accuracy
• Authenticated profile options require a declared ID to be passed at login events

Data Export Controls

• Data Export Controls (DEC) are labels applied to data sources and destinations to prevent restricted data from being sent to incompatible destinations
• Controls are set at the data source level; restrictions are set at the destination level
• If a data source control conflicts with a destination restriction, the segment is blocked from that destination
• DEC categories include: PII, on-site personalization, online behavioral advertising, and others
• Export controls apply automatically — AAM enforces them without manual review of each segment activation

Audience Marketplace and Data Governance

• Audience Marketplace allows data providers (sellers) and data buyers to transact second and third-party data
• Sellers set use case permissions: segments and overlap, modeling, and/or activation
• Buyers can only use purchased data for permitted use cases as defined by the seller
• CPM (Cost Per Mille) reporting is required from buyers — AAM auto-reports impressions for server-to-server destinations
• Flat fee plans provide unlimited access for a fixed monthly cost; CPM plans are usage-based
• Sellers can see buyer overlap reports but cannot identify individual buyer companies without consent

GDPR and Privacy Compliance in AAM

• AAM integrates with Adobe Experience Platform Privacy Service for GDPR and CCPA data subject requests
• Opt-out mechanisms include global opt-out (demdex.net opt-out), partner-level opt-out, and IAB TCF consent signals
• AAM supports IAB Transparency and Consent Framework (TCF) 2.0 for managing consent strings
• Data deletion requests remove user profiles and associated trait/segment data within the SLA timeframe
• Declared IDs can be used to submit privacy requests for known/authenticated users
• AAM's Opt-out Management tool allows architects to configure which signals honor opt-outs

Destination Types

• Cookie destinations write segment information as cookies on the user's browser for client-side activation
• URL destinations fire pixel calls when a user qualifies for or exits a segment
• Server-to-server (S2S) destinations send segment data via batch files (Amazon S3/FTP) or real-time API calls
• Adobe Experience Platform (AEP) destination allows segment sharing between AAM and AEP for unified activation
• People-Based Destinations activate hashed PII (email) to social platforms like Facebook and LinkedIn

Destination Configuration and Mapping

• Each destination requires mapping of AAM segments to partner-specific segment codes or IDs
• Serialization in cookie destinations allows multiple key-value pairs to be passed in a single cookie
• URL destinations support macro substitution (e.g., %ALIAS%, %DPUUID%) for dynamic data insertion
• Batch destinations use scheduled delivery intervals (daily by default) with configurable file formats
• Destination data export labels must be compatible with data source controls or segments are blocked
• Addressable Audience metrics at the destination level show match rates between AAM and the partner platform

Addressable Audiences and Match Rates

• Addressable Audience is the count of users who qualify for a segment AND have a synced ID with the destination
• Match rate = (Addressable Audience / Segment Population) expressed as a percentage
• Low match rates can result from ID sync gaps, cookie deletion, or cross-device profile fragmentation
• Customer Addressable Audience excludes Audience Marketplace third-party data
• Segment Addressable Audience shows unique users qualified in the last 30 days with an active ID sync
• ID synchronization frequency and recency directly impact addressable audience size and match rate quality

People-Based Destinations and Hashed IDs

• People-Based Destinations require SHA-256 hashed email addresses (hashed PII) as the cross-device identifier
• Declared ID workflow must be used to associate hashed emails with AAM device profiles
• Profile Link Device Graph must be enabled to support People-Based Destinations
• AAM normalizes (lowercases, trims) email addresses before hashing to ensure consistency
• Segments sent to People-Based Destinations are matched to platform user IDs on the destination side (e.g., Facebook Custom Audiences)

AAM Reporting and Overlap Reports

• Trait-to-Trait Overlap Report shows the unique user overlap between any two traits to identify redundancy or relationship
• Segment-to-Trait Overlap Report helps identify which traits contribute most to segment membership
• Segment-to-Segment Overlap Report reveals audience duplication across segments, useful for suppression strategies
• Overlap reports require a minimum audience size of 50,000 uniques to protect user privacy and ensure statistical validity
• General Reports provide trait, segment, and destination population data over 1-day and 30-day windows
• Trend Reports show time-series data for trait and segment populations to identify growth or decay patterns

Administration, Roles, and Multi-Tenancy

• AAM uses Role-Based Access Control (RBAC) to restrict access to data sources, traits, segments, and destinations
• Wildcard permissions allow a group to access all current and future objects of a given type
• AAM supports multi-tenancy through separate companies within a single contract (e.g., agency managing multiple brands)
• Data sources should be scoped per business unit or brand to enforce proper data governance and access control
• Audit logs are available for tracking changes to traits, segments, and destinations
• API access is controlled via OAuth 2.0 credentials configured in the Adobe Developer Console

Signal Dashboard and Unused Signals

• The Signal Dashboard shows key-value pairs received by AAM that have NOT yet been mapped to a trait
• Unused signals represent missed trait creation opportunities and should be reviewed regularly
• Top Unused Signals report is sortable by volume, helping prioritize which signals to convert into traits
• Signal search allows filtering by key name, value, source type, and date range
• High-volume unused signals often indicate implementation issues or new data sources requiring trait creation

AAM and Adobe Experience Platform (AEP) Integration

• AAM segments can be shared to AEP via the Experience Cloud Audience Sharing integration
• AEP segments created in Real-Time CDP can be pushed back to AAM for activation via AAM destinations
• The AAM-AEP connector uses the ECID namespace as the primary identity for profile stitching
• Segment sharing between AAM and AEP has a propagation latency of up to 8 hours for batch segments
• AAM is positioned as the segment activation layer while AEP handles unified profile management and data lake storage
• Organizations migrating from AAM to AEP should plan for trait/segment recreation as rule syntax differs