Audience planning and journey analytics
How Audience Agent prepares SFMC audience work for review.
Audience Agent turns campaign handoff context, scoped knowledge, saved audience definitions, and journey analytics into reviewable audience plans before Factory, Journey, or Data Extension work continues.
Workflow stages
From campaign handoff to audience and data-contract drafts.
Audience Agent is the implemented bridge between campaign planning and the audience setup details that SFMC teams need to verify: filters, inclusions, exclusions, consent requirements, fields, and entry-source readiness.
Accept campaign handoff context
Audience Agent normalizes campaignBrief, audienceDraft, journeyInfo, entrySourceDraft, buildState handoffs, and session chat history before it plans an audience or data contract.
Separate planning from analytics
The chat route classifies requests as audience definition, journey analytics, or clarification so performance questions do not get forced into audience templates.
Draft entry audience plans
The entry-audience planner returns name, description, filtersOrSql, inclusion rules, exclusion rules, data needs, consent requirements, and activationReadiness.
Prepare a data contract
The data-contract planner recommends the entry type, Data Extension name, customer key, primary key, fields, field mapping, warnings, and assumptions for review.
Journey analytics
Audience questions can use scoped performance snapshots.
When analytics data is connected, Audience Agent can answer performance questions from the account-scoped CB_Journey_Analytics snapshot instead of inventing audience size or engagement claims.
Evaluator evidence
How analytics becomes safer audience planning evidence.
Audience Agent is strongest when a pilot uses real journey performance data to narrow audience assumptions, then asks Campaign Builder to make the downstream SFMC setup explicit before anyone creates assets.
Snapshot readiness
GET /api/v1/agent/audience/analytics/status reports whether the account-scoped analytics snapshot is ready, empty, unavailable, or missing connection context before a buyer trusts performance answers.
Refreshable source data
POST /api/v1/agent/audience/analytics/refresh reloads CB_Journey_Analytics through resolved SFMC credentials and returns row count plus snapshot date evidence instead of relying on stale local assumptions.
Ambiguity handling
The chat classifier can separate analytics questions from audience-definition requests and ask which journey the user means when the analytics snapshot contains ambiguous journey references.
Planning handoff
Analytics answers can inform the next audience definition, but Data Extension creation still moves through reviewed Factory or SFMC routes rather than Audience Agent silently creating production assets.
Route-backed evidence
What evaluators can trace in the Campaign Builder implementation.
The Audience Agent page is grounded in the implemented route family under /api/v1/agent/audience and the shared audience analytics service.
Audience chat
POST /api/v1/agent/audience/chat hydrates recent session history, applies account tone guidance, grounds answers in readable account knowledge, logs requests, and can save the last assistant audience definition with /save.
Entry audience planning
POST /api/v1/agent/audience/entry-audience/plan produces the reviewed audience plan used before Journey Builder entry-source decisions.
Data contract planning
POST /api/v1/agent/audience/data-contract/plan creates Data Extension field and mapping drafts with warnings such as missing primary key or ContactKey/SubscriberKey mapping.
Analytics and saved audiences
GET /analytics/status, POST /analytics/refresh, and GET /audiences expose snapshot readiness, refresh, and saved account-scoped audience definitions.
Guardrails
Audience work stays account-scoped and reviewable.
Audience Agent helps teams define and inspect audience work, but production Marketing Cloud action still moves through authenticated, entitlement-gated, and reviewed execution paths.
Authenticated account context
Every Audience Agent API route uses requireAuth before reading account data, chat logs, saved audiences, analytics snapshots, or SFMC credentials.
Scoped knowledge and analytics
Audience replies use tenant and workspace filters for readable knowledge, account-scoped MongoDB reads, and SFMC credentials loaded for the authenticated account or user context.
Review before execution
Audience Agent prepares definitions, analytics answers, entry-audience plans, and data contracts; it does not send campaigns or activate Journey Builder journeys.
Evaluator FAQ
The boundaries to verify before a pilot.
Audience Agent is most useful when buyers understand where planning ends and reviewed SFMC execution begins.
Can Audience Agent create an audience from Campaign Agent output?
Yes. The Audience Agent route accepts campaign and audience handoff payloads, merges them with buildState context, and treats that first-party handoff as source material for the audience definition.
What happens when a request mixes analytics and audience planning?
The chat classifier can ask a clarification question instead of guessing. It distinguishes audience-definition requests from journey performance questions and can ask which journey the user means when a reference is ambiguous.
Does Audience Agent create Data Extensions directly?
Audience Agent prepares entry-audience and data-contract plans. Factory and SFMC Data Extension routes handle reviewed creation work when the account has the required credentials, configuration, and entitlements.
Does Audience Agent activate journeys or approve sends?
No. It supports audience planning and analytics review. Journey activation, send approval, and final Marketing Cloud launch decisions remain manual reviewed steps.
Evaluation takeaway
Use Audience Agent to make audience setup explicit before SFMC build work.
A practical pilot should bring a real campaign brief, a target audience hypothesis, the expected Journey Builder entry source, and any Data Extension field requirements. Audience Agent can turn that context into reviewable audience and data-contract drafts before Factory or Journey work proceeds.