SFMC pilot plan
A practical Campaign Builder pilot plan for Salesforce Marketing Cloud teams.
Use one real campaign workflow to validate the implemented path from authenticated SFMC context to Campaign Agent planning, Bruno dry-run orchestration, specialist agent work, Factory readiness, and reviewed Marketing Cloud drafts.
Pilot sequence
Validate the same controls the product uses in production.
Campaign Builder is most credible when the pilot follows the implemented route families: auth and BU mapping first, scoped entitlements second, SFMC credential checks third, then Campaign Agent, Bruno, specialist agents, and Factory where configured.
Prove the launch context
Start with the actual entry path: first-party cb_auth session, bearer JWT, SFMC signed_request launch, or SFMC mc_token session. For SFMC app launches, confirm the enterprise ID and business unit ID resolve to the intended Campaign Builder tenant and workspace before agent or Factory routes run.
Limit the capability surface
Enable only the pilot workflows the account should test. Campaign, Content, Email, Journey, Image, Translator, Advisor, Audience, Bruno, content-blocks, and tool:factory are separate entitlement gates, so the pilot can expose a controlled scope.
Connect SFMC credentials
Use account-scoped server-to-server or web-app Marketing Cloud credentials, then verify token context before Content Builder assets, email assets, journeys, folders, Data Extensions, or Factory routes are evaluated.
Run one campaign through planning
Use Campaign Agent to turn typed intent, uploaded files, session context, and workspace knowledge into a normalized brief, audience direction, email content plan, Data Extension plan, journey plan, validation warnings, build package, and Bruno-compatible campaignSpec.
Dry-run Bruno before writes
Bruno should discover entitled skills, collect missing inputs, show pending write actions with preview summaries, and produce a dry-run stage plan without SFMC side effects before any confirmed execution.
Review draft outputs in SFMC
Validate the reviewed paths that matter for the pilot: email copy or update, slot updates, translation, Content Builder publishing, image insertion, Factory drafts, Data Extensions, previews, and Journey Builder drafts with manual activation still inside Marketing Cloud.
Evidence checklist
What to capture during the first evaluation run.
The best pilot output is a set of traceable artifacts that SFMC admins, campaign operators, and security reviewers can inspect together.
Authentication evidence
Record the session source, CSRF behavior for same-origin writes, requestId on failures, and the normalized account, tenant, workspace, enterprise, and business-unit context.
Entitlement evidence
Show that unavailable agents and tools stay unavailable. The pilot should not expose Bruno, Factory, Email Agent, Journey Agent, or other specialist surfaces unless the matching entitlement is enabled.
Campaign planning evidence
Keep the generated brief, build package, validation warnings, grounding references, Data Extension plan, journey plan, email definitions, and campaignSpec from Campaign Agent.
Bruno orchestration evidence
Capture dry-run output, pending action summaries, durable run state for confirmed execution, stage status, artifact ledger, request journal, resume behavior, and manual activation warnings.
Factory readiness evidence
Verify tool:factory entitlement, settings.factory.enabled set to true, connected SFMC credentials, configured template assets, template checks, campaign drafts, folder preparation, Data Extension creation paths, email previews, variants, and template review.
SFMC write-boundary evidence
Confirm production-bound work is previewed and confirmed before it reaches Content Builder, Email Studio assets, Data Extensions, image insertion, Factory routes, or Bruno write-skill execution.
Pilot artifact matrix
Tie every pilot claim to a route, artifact, and boundary.
A buyer should be able to review what was run, what evidence was produced, and which Marketing Cloud control stayed outside Campaign Builder automation.
Launch and connection checkpoint
Implemented surface: /api/v1/sfmc/app/launch, /api/v1/sfmc/oauth/callback, /api/v1/sfmc/bu-map, and /api/v1/sfmc-connect/test.
Evidence to keep: Capture the auth source, mapped enterprise and business-unit IDs, tenant/workspace IDs, connection status, token-context response, and requestId on errors.
Boundary to verify: A valid SFMC connection proves identity and credential readiness; it does not bypass entitlement gates or confirm any production write.
Campaign planning checkpoint
Implemented surface: /api/v1/agent/campaign/generate-brief and /api/v1/agent/campaign/execute-campaign.
Evidence to keep: Keep the normalized brief, uploaded-context references, buildPackage, Data Extension plan, journey plan, validation warnings, and Bruno-compatible campaignSpec.
Boundary to verify: Campaign Agent prepares the execution package. Confirmed execution still routes through Bruno and creates draft artifacts only.
Bruno dry-run checkpoint
Implemented surface: /api/v1/agent/bruno/execute-campaign and Bruno chat pending-action state.
Evidence to keep: Capture dryRun output, stage plan, pending actions, previewSummary text, durable run state, artifact ledger, request journal, and resume behavior if a run is interrupted.
Boundary to verify: Dry runs must show proposed work without SFMC side effects. Write skills stay confirmation-gated before execution.
Factory production checkpoint
Implemented surface: /api/v1/factory/campaigns/draft, /api/v1/factory/assets/folder, /api/v1/factory/data-extensions, /api/v1/factory/emails, /api/v1/factory/preview/draft, /api/v1/factory/preview/variants, and /api/v1/factory/template-review.
Evidence to keep: Verify tool:factory entitlement, settings.factory.enabled, template-check results, campaign draft records, folder IDs, Data Extension outputs, email previews, variant previews, and template review findings.
Boundary to verify: Factory prepares configured SFMC drafts and supporting assets; it should not be evaluated before credentials, templates, and review paths are ready.
Journey draft checkpoint
Implemented surface: /api/v1/marketingcloud/createJourneyFromScratch and /api/v1/marketingcloud/createJourneyFromTemplate.
Evidence to keep: Save the journeyDraftPlan, selected source journey or scratch template, created email references, activationReadiness, and any cloneArtifacts warnings such as EMAIL_LEGACY_ID_MISSING or ENTRY_EVENT_CREATE_FAILED.
Boundary to verify: Journey creation is draft-only. Review, entry-source repair, EMAILV2 selection, and activation remain manual Marketing Cloud steps where required.
Pilot runbook
Run the pilot as a controlled SFMC execution rehearsal.
Treat the first campaign as a production-style rehearsal with named owners, route-backed actions, and evidence that separates Campaign Builder drafts from Marketing Cloud approval and activation.
Prep the workspace
Owner: SFMC admin and Campaign Builder admin
Action: Confirm the launch path, BU-to-workspace mapping, account-scoped credentials, token context, route entitlements, `settings.factory.enabled` if Factory is in scope, and configured template assets before marketers start the campaign run.
Evidence: Keep screenshots or API responses for `/api/v1/sfmc/app/launch`, `/api/v1/sfmc/bu-map`, `/api/v1/sfmc-connect/test`, entitlement settings, Factory template-check output, and request IDs on failed checks.
Build the campaign package
Owner: Campaign operator
Action: Use Campaign Agent with the real campaign idea, uploaded context, workspace knowledge, audience direction, messaging needs, journey direction, and Data Extension needs until the build package and campaignSpec are reviewable.
Evidence: Save the normalized brief, grounding references, assumptions, validation warnings, email definitions, Data Extension plan, journey plan, buildPackage, and Bruno-compatible campaignSpec.
Dry-run orchestration
Owner: Campaign operator and SFMC admin
Action: Run Bruno with dryRun enabled, then inspect entitled skills, missing inputs, proposed folders, Data Extensions, reused assets, email drafts, preview drafts, stage warnings, and manual Journey activation warnings before any confirmed execution.
Evidence: Capture the Bruno dry-run response, pending action preview summaries, stage plan, proposed artifact list, manual activation warnings, and any missing prerequisites the team must resolve.
Produce reviewed drafts
Owner: Campaign operator
Action: Confirm only the write paths selected for the pilot: Email Agent copy or update, Content Builder save, image insertion, Factory folders, Data Extensions, template-based emails, preview drafts, variant previews, or Journey draft creation.
Evidence: Record created or reused SFMC asset IDs, folder IDs, Data Extension customer keys, email preview links, Factory preview and template-review responses, Bruno run journal entries, and artifact ledger rows.
Review in Marketing Cloud
Owner: SFMC admin and campaign approver
Action: Open the draft artifacts in Salesforce Marketing Cloud, verify content, data setup, entry-source readiness, EMAILV2 warnings, send approval requirements, and manual Journey activation steps outside Campaign Builder.
Evidence: Keep the final go or no-go notes, manual SFMC fixes, unresolved blockers, activationReadiness state, request IDs, and the list of actions that remained manual in Marketing Cloud.
Workflow tests
Use the campaign to test concrete implemented workflows.
Pick the workflow paths that match the pilot campaign. Each one maps to real Campaign Builder agent routes or Factory routes rather than unsupported launch automation.
Campaign idea to structured brief
Use Campaign Agent generate-brief and build flows to prove uploaded context, workspace knowledge, assumptions, naming, validation warnings, and Bruno-ready campaignSpec output.
Existing email reuse
Use Email Agent search, preview, copy, rewrite, translate, or slot-aware update paths against accessible SFMC email assets, with preview summaries instead of raw hidden HTML writes.
Journey draft boundary
Use Journey Agent to extract a brief, search or clone a reference journey, prepare email and audience artifacts, run readiness checks, and create a draft while keeping activation manual.
Factory production setup
Use Factory to save campaign drafts, create Content Builder folders, upload images, prepare Data Extensions, generate template-based emails, request previews, review variants, and run template review.
90-minute evaluation agenda
Keep the first buyer session focused on inspectable SFMC evidence.
This agenda gives SFMC admins and campaign operators a practical walkthrough that follows Campaign Builder's implemented execution order instead of jumping straight from chat output to production-bound writes.
0-30 minutes
Prove SFMC identity and scope
Start from the real launch path and capture the mapped enterprise ID, business unit ID, tenant ID, workspace ID, account-scoped credential status, token-context result, entitlement scope, and requestId behavior before any agent route is evaluated.
30-60 minutes
Build and dry-run one campaign
Use Campaign Agent to produce the normalized brief, grounding references, validation warnings, Data Extension plan, journey plan, email definitions, buildPackage, and Bruno-compatible campaignSpec, then run Bruno with dryRun enabled to inspect missing inputs, proposed stages, pending write previews, and no SFMC side effects.
60-90 minutes
Confirm only reviewed draft outputs
If prerequisites are ready, confirm the selected Email Agent, Content Builder, Image Agent, Factory, Data Extension, preview, or Journey draft path, then verify the created or updated SFMC artifact IDs while keeping send approval and Journey activation as manual Marketing Cloud decisions.
Review boundaries
Keep the pilot inside supported SFMC execution boundaries.
Campaign Builder helps teams create and review SFMC draft artifacts. The pilot should make that boundary explicit instead of implying automatic sends, unrestricted writes, or certification claims.
Do not treat a dry run as a send or Journey activation.
Do not enable every agent if the pilot only needs one campaign workflow.
Do not evaluate Factory before SFMC credentials, settings.factory.enabled, tool:factory entitlement, and template assets are ready.
Do not count unchecked chat output as production evidence; use preview summaries, route responses, request IDs, and SFMC draft artifacts.
Do not position Campaign Builder as bypassing Marketing Cloud review, approval, or activation controls.
FAQ
Questions to settle before confirming SFMC writes.
What is the smallest meaningful Campaign Builder pilot?
One real campaign idea is enough if it exercises the implemented path: authenticated SFMC context, Campaign Agent brief and build package, Bruno dry run, selected specialist agent work, and reviewed SFMC draft outputs or explicit readiness blockers.
Should Factory be part of every pilot?
No. Include Factory when the evaluation needs production draft workflows for campaign drafts, folders, image uploads, Data Extensions, template-based emails, previews, variants, or template review. Otherwise validate the agent and Bruno flows first.
What should an SFMC admin verify first?
Verify auth mode, BU-to-workspace mapping, account-scoped Marketing Cloud credentials, entitlement scope, token context, Content Builder access, Journey Builder access, Data Extension permissions, and the manual Journey activation boundary.
What counts as a successful dry run?
A useful dry run shows the proposed stage plan and artifacts without SFMC side effects, identifies missing inputs, lists pending write actions with preview summaries, and keeps manual activation warnings visible.
Next reading
Pair the pilot plan with the technical pages.
Use the end-to-end execution article, SFMC connection readiness guide, Bruno article, Factory workflow, and technical FAQ as the supporting detail for each pilot checkpoint.