Data 360 Implementation Guide

This guide sets out a phased plan for implementing Data 360, formerly Data Cloud, with a checklist for each phase. It is about sequencing and decisions: what to settle first, what is hard to change later, and what to check before moving on. For how the layers fit together, read the Data 360 architecture guide first; this guide assumes that picture.

Facts about Data 360 behaviour below come from Salesforce documentation and Trailhead. Features and limits change between releases, so confirm details in Salesforce Help for your org and edition.

Phase 0: Readiness

Before any configuration, confirm the use cases and the data they need. Our Data 360 readiness checklist and readiness self-assessment cover this in detail.

• Three to five priority use cases, each with a named business owner and a measurable outcome.
• The freshness each use case needs: seconds, minutes, hours or days.
• A source inventory with owners, volumes and identifiers.
• The current consent position for each channel.
• Licences confirmed: Data 360 provisioned, and any dependent products such as Marketing Cloud Next or Salesforce Personalization.

Phase 1: Architecture and data flow

• Draw the end-to-end flow for each use case, from source to destination. Our Data 360 reference architecture is a starting point.
• Mark where data leaves your Salesforce trust boundary: activation targets, webhooks and middleware.
• Decide where zero copy federation replaces ingestion. Salesforce lists out-of-the-box support for Snowflake, Databricks, Redshift and BigQuery; file federation is labelled "In Beta Now" on its page, so check Help.
• Choose the timing path per use case. Standard web and mobile streams ingest engagement data every 15 minutes and profile data hourly; the real-time layer targets about 500 ms for 95% of events and needs real-time data graphs.

Phase 2: Sources and streams

• Use Salesforce's "zero cost connectors" for Sales, Service, Industries, B2C Commerce and Marketing Cloud Engagement where they fit.
• For each stream, set the category (Profile, Engagement or Other), the primary key and, for Engagement, the Event Time field.
• Decide the refresh mode deliberately: Salesforce states it "cannot be changed" after the stream is created.
• Allow for the Ingestion API taking up to an hour to show data.
• See choosing data stream ingestion patterns.

Phase 3: Modelling approach

• Map DLOs to standard DMOs first; add custom DMOs or fields only where the standard model has no home.
• Use starter data bundles to map standard Salesforce sources.
• Use stream formula fields to build missing keys or flags.
• Record every mapping in a data dictionary.
• See mapping custom sources to the Customer 360 Data Model.

Phase 4: Identity and matching

• Choose match rules (exact, fuzzy, normalised) per identifier, for example Fuzzy Name plus Normalised Email or Phone.
• Choose reconciliation rules per attribute: most frequent, most recent or source priority.
• Prepare known duplicate sets to test against.
• Decide whether runs are automatic. Salesforce notes that turning off "Run jobs automatically" can reduce cost.
• Real-time identity resolution needs a real-time data graph.

Phase 5: Data quality

• Profile each source for completeness, format and duplicates before you tune match rules.
• Normalise email and phone formats at source or in a transform.
• Agree thresholds and owners for each quality measure.
• See data quality checks before identity resolution.

Phase 6: Permissions and consent

• Define roles: source admin, modeller, segment builder, activation manager, analyst.
• Build least-privilege permission sets for each.
• For Marketing Cloud Next, model consent with the Communication Subscription Consent objects, which Salesforce stores as DMOs and checks at send time.
• Confirm the Interactions SDK consent behaviour on your sites.
• See Data 360 governance.

Phase 7: Transformation strategy

• Use batch transforms (scheduled, visual editor, more functions) for heavier reshaping and joins.
• Use streaming transforms (continuous, SQL-based) for light, record-by-record changes.
• Keep transform logic documented and owned; it is easy to lose track of.

Phase 8: Activation

• Confirm each destination is a supported activation target in current Help; design an integration path where it is not.
• Activate only the attributes each destination needs.
• Use data actions to Platform Events, Marketing Cloud or a webhook for event-like triggers.
• See Data 360 activation.

Phase 9: Monitoring and failure handling

• Monitor stream status, identity resolution jobs, segment publishes and activations.
• Alert on failures and on unusual swings in profile or segment counts.
• Write a runbook for each job: what failure looks like, who responds and how to recover.
• See monitoring Data 360 jobs and refresh schedules.

Phase 11: Licensing and capacity

• Model processing consumption before go-live. Salesforce's Agentforce ROI calculator notes that it excludes Data 360 credit costs, so treat them as a separate line.
• Real-time features run continuously; scope real-time data graphs to the attributes decisions need.
• If you use the hosted Data 360 MCP server, its calls count toward Connect API limits and Flex Credits.
• Check limits and pricing with your Salesforce account team; this guide quotes none.

Phase 12: Deployment

• Build and test in a sandbox or test environment before production.
• Document every configuration item, because some Data 360 setup may need to be recreated rather than deployed. Check Salesforce Help for the current options for moving configuration between environments.
• Release in a fixed order: streams, mappings, transforms, identity, insights, segments, activations.
• Plan the first identity resolution run and first publishes outside busy campaign windows.
• See DevOps and release management and the release readiness checklist.

After go-live

Review costs, job health and use case outcomes monthly for the first quarter. Add sources one at a time, each with an owner and a test plan. Unannounced source changes can break mappings, identity rules and segments, so agree a change notice process with source owners.

Sources

• https://developer.salesforce.com/docs/data/data-cloud-int/guide/c360-a-create-ingestion-data-stream.html
• https://trailhead.salesforce.com/content/learn/modules/data-cloud-connect-and-unify/connect-and-transform-data
• https://trailhead.salesforce.com/content/learn/projects/explore-data-cloud-core-functionality/unify-your-data
• https://trailhead.salesforce.com/content/learn/modules/batch-data-transforms-in-data-cloud-quick-look/get-started-with-batch-data-transforms-in-data-360
• https://trailhead.salesforce.com/content/learn/projects/explore-data-cloud-core-functionality/build-a-segment-and-report
• https://www.salesforce.com/agentforce/ai-agents-roi-report/