Campaign Planning And Attribution

This skill activates when a practitioner needs to design Campaign Hierarchy structures for program-level ROI aggregation, configure Customizable Campaign Influence (CCI) attribution models in Salesforce, or interpret multi-touch revenue attribution data from MCAE B2B Marketing Analytics. It covers architecture decisions and configuration — not execution of individual campaigns or MCAE connector provisioning.

---

Before Starting

Gather this context before working on anything in this domain:

Check	What the platform actually says
Is Customizable Campaign Influence on?	CampaignSettings.enableCampaignInfluence2 — "Indicates whether Customizable Campaign Influence is enabled… The default value is true." Do not assume it needs turning on; retrieve settings/Campaign.settings-meta.xml and read it
Is Campaign Influence 1.0 still in play?	It cannot be, if CCI is on: "When true, Campaign Influence 1.0 is hidden from users and is no longer active", and enableAutoCampInfluenceDisabled states "enableCampaignInfluence2 must be false to use this setting"
Can the user even see Campaigns?	Campaign and CampaignMemberStatus are "defined only for those organizations that have the marketing feature enabled and valid marketing licenses… accessible only to those users that are enabled as marketing users" — otherwise they are absent from describeGlobal() entirely
Which fields carry hierarchy totals?	Not ActualCost / AmountWonOpportunities. Those are single-campaign fields. The hierarchy totals are HierarchyActualCost, HierarchyAmountWonOpportunities, TotalNumberofResponses and their siblings
How deep can the hierarchy go?	UNVERIFIED (2026-09-05): no maximum Campaign hierarchy depth appears in the Object Reference, the Metadata API guide, or the App Limits cheat sheet (grep -i campaign over the cheat sheet returns zero hits). The commonly-quoted "5 levels" has no source here — confirm in the org, do not design to a number you cannot cite
Does an external system publish models?	enableB2bmaCampaignInfluence2 — "whether your org can access campaign influence models from other systems, such as Pardot"; requires enableCampaignInfluence2 = true

Sources for the quotes above: api_meta.txt:111556–111571, object_reference.txt:57860–57863, object_reference.txt:57273–57359, object_reference.txt:57708–57802, object_reference.txt:58672–58675. Full line-by-line grounding is in references/worked-examples.md.

---

Questions to Ask Before Configuring

Ask	Why it matters	What a good answer adds
"Which single question does the attribution number have to answer — who to pay, where to spend next, or what to report to the board?"	Only one model can be the default, and only the default model's records appear on campaigns and opportunities. Everything else is report-only	The name of the default model, and an explicit statement that the others are reporting artefacts
"Are campaigns run per region, per product line and per quarter at the same time?"	Each dimension someone wants to slice by is either a hierarchy level or a Campaign field, and hierarchy headroom is the resource you cannot cite a limit for	The dimension list, split into "hierarchy" and "custom field on Campaign"
"Who writes Campaign Member records — a person, a data load, or Account Engagement?"	An invalid Status is silently coerced to the campaign's default status, not rejected. A loader is the one writer that will never notice	The writer list, and whether each writer sends the status text (required) or an Id (silently wrong)
"For each campaign type, which status means 'they responded' to the business?"	HasResponded on the status is what NumberOfResponses and HierarchyNumberOfResponses count. Since API 39.0 at least one status per campaign must have it	The per-type status table with exactly one default and at least one responded flag
"Will anything other than the recalculation job write CampaignInfluence rows?"	A locked model can only be written via the API; rows written against the Primary Campaign Source model are deleted on recalculation	The writer for each model, and a decision on isModelLocked per model
"Is anyone allowed to deactivate a model once the quarter's reporting is out?"	Deactivating a model deletes its campaign influence records. There is no undo, and reactivating does not restore them	A named owner for model lifecycle and a change-control note in the release calendar
"Which reports and dashboards will read this, and who runs them?"	Campaign visibility is licence- and marketing-user-gated before sharing is even considered; a report that returns nothing is often an access problem	The consuming report list with the running user's licence and marketing-user flag confirmed

What a proper configuration adds over just enabling Campaign Influence: the default model is a deliberate choice rather than whatever shipped, member statuses cannot silently coerce a load into the wrong funnel stage, hierarchy totals are read from the fields that actually hold them, and deactivating a model is a governed change rather than an irreversible data deletion someone discovers a quarter later.

---

Core Concepts

Campaign Hierarchy and where the totals actually live

Campaigns nest through ParentId, a lookup to Campaign (ParentCampaign is the read-only label for "the campaign above the selected campaign in the campaign hierarchy"). The rollup is not a Master-Detail roll-up summary you configure — it is a set of pre-built calculated fields with distinct names.

Scope	Field
This campaign only	ActualCost, BudgetedCost, ExpectedRevenue, NumberOfLeads, NumberOfResponses, AmountAllOpportunities, AmountWonOpportunities
The whole hierarchy	HierarchyActualCost, HierarchyBudgetedCost, HierarchyExpectedRevenue, HierarchyNumberOfLeads, HierarchyNumberOfResponses, HierarchyAmountAllOpportunities, HierarchyAmountWonOpportunities
The whole hierarchy, second family	TotalAmountAllOpportunities, TotalAmountAllWonOpportunities, TotalNumberofLeads, TotalNumberofOpportunities, TotalNumberofResponses, TotalNumberofWonOpportunities

HierarchyNumberOfResponses counts "contacts and unconverted leads with a Member Status equivalent to 'Responded' for the campaign in a campaign hierarchy" (object_reference.txt:57353–57359). Two independently named families cover the same idea; pick one per report and say which in the report description, because a dashboard mixing Hierarchy* and Total* columns will look inconsistent for reasons no one can see.

Customizable Campaign Influence

CampaignInfluence is "the association between a campaign and an opportunity in Customizable Campaign Influence", available in API 37.0 and later, and the Object Reference is explicit that it "applies only to Customizable Campaign Influence and not to Campaign Influence 1.0". Its fields are CampaignId, CampaignMemberId, ContactId, Influence (percent), ModelId, OpportunityContactRoleId, OpportunityId and RevenueShare — there is no field called Revenue.

CampaignInfluenceModel is read-only as an sObject (describeSObjects(), query(), retrieve() only) but is addable as a metadata component. Its ModelType picklist is a closed set:

Value	Model
1	Primary Campaign Source Model
2	Custom Model
3	First Touch Model
4	Last Touch Model
5	Even Distribution Model
6	Data-Driven Model

The metadata type carries no modelType element, so a model you deploy arrives as a custom model. What you do control on deploy is isActive, isDefaultModel, isModelLocked, modelDescription, name and recordPreference:

<?xml version="1.0" encoding="UTF-8"?>
<CampaignInfluenceModel xmlns="http://soap.sforce.com/2006/04/metadata">
    <isActive>true</isActive>
    <isDefaultModel>false</isDefaultModel>
    <isModelLocked>true</isModelLocked>
    <modelDescription>100% influence attribution to the last campaign that touched the contact.</modelDescription>
    <name>Acme Last Touch</name>
    <recordPreference>RecordsWithAttribution</recordPreference>
</CampaignInfluenceModel>

recordPreference is the switch most designs get wrong: AllRecords "creates records regardless of the revenue attribution percentage"; RecordsWithAttribution "creates records only when the revenue attribution is greater than 0%". Choosing AllRecords on a high-volume org multiplies row count by every touched campaign, including the ones credited nothing.

UNVERIFIED (2026-09-05): the claim that CCI requires Opportunity Contact Roles is not stated in any of the eight official extracts. What is documented is that CampaignInfluence carries ContactId and OpportunityContactRoleId fields (object_reference.txt:57941–57946, 57963–57969), which makes contact-role population a strong practical prerequisite but not a quoted platform rule. Verify in the org before presenting it as a requirement.

Campaign Member Status and funnel tracking

Two objects share the name. The org-wide default list is a StandardValueSet whose fullName is CampaignMemberStatus, mapping to CampaignMember.Status (api_meta.txt:141856). The per-campaign list is the CampaignMemberStatus sObject — CampaignId, Label (765 characters), IsDefault, HasResponded, SortOrder — and it is not a metadata type at all.

CampaignMember.HasResponded is read-only; Status controls it. Since API version 39.0, every campaign must have a default status and at least one status with HasResponded = true. A status that is the default, or is in use on a campaign, cannot be deleted.

MCAE Multi-Touch Attribution App

UNVERIFIED (2026-09-05): time-decay, position-based / U-shaped models, the "Multi-Touch Attribution App", and any weighting percentages for them appear nowhere in the Metadata API guide, the Object Reference, or the other six extracts (grep -i "time.decay\|u-shaped\|position-based\|multi-touch" returns no hits). The only related grounded facts are CampaignSettings.enableB2bmaCampaignInfluence2 ("whether your org can access campaign influence models from other systems, such as Pardot") and PardotSettings.enableEnhancedProspectCustomFieldsSync ("Enable Object Sync to enhance with B2B Marketing Analytics or B2B Marketing Analytics Plus"). Treat model names and weightings from Account Engagement as vendor documentation to confirm in the org, and never present a 40/40/20 split as a platform fact. admin/mcae-pardot-setup owns the connector itself.

---

Common Patterns

Pattern: Program/Tactic Hierarchy with Rolled-Up ROI Dashboard

When to use: A marketing team runs multiple channels (email, webinar, paid) under a single program and needs a unified ROI view at the program level for executive reporting.

How it works:

1. Create a parent Campaign at the program level. Set BudgetedCost and ExpectedRevenue here.
2. Create child Campaigns (Type = Email, Webinar, Advertisement) and link them via ParentId.
3. Populate ActualCost on each child campaign as spend is confirmed.
4. Build the program dashboard on HierarchyActualCost and HierarchyAmountWonOpportunities on the parent — not on ActualCost, which only ever holds the parent's own spend.
5. State the field family used in the report description so a second author does not mix Hierarchy* with Total*.

Why not the alternative: Without hierarchy, teams manually aggregate per-channel costs into a spreadsheet. This breaks when campaigns are added mid-program and creates reconciliation debt.

Pattern: Dual-model CCI — one default, one reporting-only

When to use: Revenue operations needs first-touch and last-touch views side by side.

How it works:

1. Retrieve settings/Campaign.settings-meta.xml and confirm enableCampaignInfluence2 is true.
2. Deploy two CampaignInfluenceModel files. Exactly one carries isDefaultModel = true; that one drives the Campaign Influence related list on opportunities, the Influenced Opportunities related list on campaigns, and the Campaign Statistics section.
3. Set isModelLocked = true on both so only the API writes rows, then name the writer for each.
4. Set recordPreference = RecordsWithAttribution unless zero-credit rows are genuinely wanted.
5. Build one custom report type on CampaignInfluence joined to Opportunity and group by ModelId — one report type serves both models.

Why not the alternative: Two report types, one per model, doubles the maintenance for no gain: ModelId is a column, not a schema difference.

---

Decision Guidance

Situation	Recommended Approach	Reason
Need program-level ROI across multiple campaign tactics	Campaign Hierarchy, dashboard on Hierarchy* fields	The Hierarchy* and Total* families are the only fields that aggregate descendants
Need multiple campaigns to share credit on one opportunity	CCI with one custom model per view	Only CCI produces multiple CampaignInfluence rows per opportunity per model
Need time-decay or U-shaped models	Confirm the capability in the org first, then MCAE / B2BMA Plus	UNVERIFIED (2026-09-05): those model names are absent from the official extracts; ModelType documents six values and none of them is time-decay
Someone asks to "turn off Campaign Influence 1.0 first"	Check enableCampaignInfluence2 — if true, 1.0 is already hidden and inactive	enableAutoCampInfluenceDisabled is only usable while enableCampaignInfluence2 is false
A model's rows are wrong and someone wants to "reset" it	Never by deactivating	Deactivating a model deletes its campaign influence records
Attribution rows exist but the related lists are empty	Check which model is default	Only the default model's records appear on campaigns and opportunities
Need real-time campaign spend reporting	Report on child Campaign ActualCost directly	The Hierarchy* fields are documented as calculated; no refresh SLA is published

---

Recommended Workflow

1. Retrieve before designing — sf project retrieve start --metadata CampaignInfluenceModel --metadata "Settings:Campaign" --metadata "CustomObject:Campaign". Read enableCampaignInfluence2, enableB2bmaCampaignInfluence2 and enableAccountsAsCM off the retrieved file rather than asking. Confirm the running user is a marketing user with a marketing licence, or Campaign will not describe at all.
2. Answer the seven questions above and fill templates/campaign-planning-and-attribution-template.md — in particular the default-model choice, the per-type status table, and the model-lifecycle owner.
3. Write the campaign plan record — the YAML artefact in references/worked-examples.md § 1: hierarchy, naming regex, member status sets, attribution models. This is the single source the rest of the build is generated from.
4. Lint it — python3 scripts/check_campaign_planning_and_attribution.py --file <plan>.yaml --manifest-dir force-app/main/default. It fails on a duplicate campaign id, a name that breaks the naming regex, a status set without exactly one default or without a responded status, a model_type outside the six documented values, more than one is_default_model, a bad recordPreference enum, and a record type whose Type picklist values are not in the plan.
5. Generate and deploy the metadata — CampaignSettings, the CampaignInfluenceModel files, Campaign fields and record type, the two standard value sets, and the report type, using the package.xml in references/worked-examples.md § 7 (remember: settings are declared under <name>Settings</name>, and RecordType and StandardValueSet do not accept *).
6. Create the per-campaign member statuses — these are sObject rows, not metadata. Run the member-status health query in § 9 afterwards to catch any campaign left without exactly one default.
7. Run the acceptance tests in § 10 — especially A4 (deactivation deletes rows), A5 (invalid status coerces to default), and A7 (Lead + Contact on one member inserts silently as Contact only). Record the results in the template before handing over.

---

Review Checklist

• enableCampaignInfluence2 read from the retrieved settings file, not assumed
• Exactly one CampaignInfluenceModel has isDefaultModel = true
• recordPreference is a deliberate choice per model, not a copied default
• isModelLocked matches the named writer for each model
• Dashboards read Hierarchy* / Total* fields, never ActualCost as a rollup
• One field family (Hierarchy* or Total*) per report, stated in the description
• Every campaign has exactly one default member status and at least one responded status
• Every writer of CampaignMember sends the status text, never the status Id
• Model deactivation has a named owner and a change-control entry
• Report consumers are marketing users with a marketing licence
• scripts/check_campaign_planning_and_attribution.py exits 0 on the plan and the manifest

---

Salesforce-Specific Gotchas

The nine behaviours behind most campaign-attribution incidents are in references/gotchas.md, each with what happens, when it occurs, how to avoid it, and its guide line range. The three that cost the most:

#	Behaviour
1	Deactivating an influence model deletes its rows; reactivating does not bring them back
2	An invalid CampaignMember.Status is coerced to the campaign's default status and inserts cleanly — nothing errors
3	ActualCost on a parent campaign is that campaign's own spend; the hierarchy total is HierarchyActualCost

---

Output Artifacts

Artifact	Description
Campaign plan record (YAML)	Hierarchy, naming convention, per-type member status sets, attribution models — the input to the checker
Campaign.settings-meta.xml	The org switches, retrieved and re-deployed deliberately
*.campaignInfluenceModel-meta.xml	One file per model, with the default and lock decisions recorded
Campaign fields + record type	The dimensions kept off the hierarchy
*.reportType-meta.xml	CampaignInfluence joined to Opportunity, grouped by model
Acceptance test results	The ten Given/When/Then rows, run in the UAT sandbox

---

Reference Files

File	Read it when
references/worked-examples.md	Building anything — the Acme quarter carried end to end: the plan YAML, CampaignSettings, three CampaignInfluenceModel files, Campaign fields and record type, both standard value sets, the report type, package.xml with its four wildcard answers, the retrieve/deploy commands, the ROI and attribution SOQL, and ten acceptance tests
references/gotchas.md	Nine platform behaviours behind most attribution incidents — model deactivation deleting rows, status coercion, the two rollup field families, Lead-XOR-Contact, Primary Campaign Source rows being recalculated away
references/examples.md	Three worked designs — hierarchy ROI, dual-model CCI, and the member-status repair — plus the anti-pattern of automating on a calculated campaign field
references/well-architected.md	Pillar mapping, the architectural tradeoffs, and the official-source list behind every claim in this package
references/llm-anti-patterns.md	Self-checking generated output — the ways an assistant gets campaign attribution wrong
templates/campaign-planning-and-attribution-template.md	Capturing the design before building it: hierarchy table, status sets, model decisions, and the manual-Setup register

---

Related Skills

• admin/mcae-pardot-setup — MCAE connector provisioning, business units, and the campaign sync that feeds attribution data into Salesforce
• admin/marketing-reporting-requirements — agreeing which attribution model and which KPIs before any configuration exists
• admin/opportunity-management — Opportunity.CampaignId (the primary campaign source), contact roles, stages and splits
• admin/lead-management-and-conversion — what conversion does to campaign membership and lead-sourced attribution
• admin/report-type-strategy — choosing the base object and join shape for the influence report type
• admin/reports-and-dashboards — building the report and dashboard on top, folder sharing, and the running user