4.2 Auditing, Status Monitoring, Error Logs & Data Explorer
Key Takeaways
- Data Stream status indicators—Active, Inactive, Error, and Paused—reflect ingestion health, with the Refresh History tab providing run-level execution logs, record count tallies, and downloadable error manifests.
- Data Explorer enables administrators and Data Aware Specialists to directly inspect and query raw Data Lake Objects (DLOs), canonical Data Model Objects (DMOs), and Calculated Insights, validating mappings, transformations, and schema fidelity up to 100 sample records per view.
- Profile Explorer is the dedicated interface for inspecting unified customer records, allowing users to search by Contact Point (email, phone, party ID) to verify Identity Resolution consolidation, reconciled attribute values, and chronological engagement timelines.
- Data Cloud usage is metered through Data Cloud Credits (DCC), with consumption driven by four primary engines: Ingestion (streaming events and batch data volumes), Data Processing (batch transforms and calculated insights), Identity Resolution graph runs, and Segmentation/Activation processing.
- Ingestion failures are predominantly caused by schema mismatches (unexpected or missing columns), datetime parsing errors (non-ISO 8601 UTC formats), and primary key collisions in batch files.
Auditing, Status Monitoring, Error Logs & Data Explorer
Operating an enterprise customer data platform requires rigorous operational monitoring, ongoing data quality auditing, and transparent consumption management. Because Salesforce Data Cloud continuously ingests massive volumes of streaming and batch data from diverse external systems, transforms it into canonical models, and calculates real-time insights, system administrators and consultants must possess deep diagnostic expertise.
When data pipelines stall, records fail to map, or identity profiles fail to merge as expected, consultants must systematically leverage Data Stream Status Logs, Data Explorer, and Profile Explorer to isolate root causes and safeguard downstream activations.
Monitoring Data Stream Health & Run History
The Data Streams tab serves as the operational nerve center for ingestion monitoring. Every configured stream—whether ingesting from Salesforce CRM, Amazon S3, Google Cloud Storage, Web & Mobile SDKs, or the Ingestion API—maintains real-time status indicators and detailed execution histories.
Data Stream Status Indicators
Data Cloud displays distinct status badges on each stream record to communicate its operational state:
| Stream Status | Operational Meaning | System Behavior & Ingestion State |
|---|---|---|
| Active | Healthy & Scheduled | Stream is operational; batch schedules or streaming listeners are processing records normally. |
| Inactive | Disabled / Not Started | Stream is intentionally disabled by an administrator or has not completed initial deployment. |
| In Progress / Processing | Job Currently Executing | Ingestion engine is actively reading source files, staging data into lakehouse storage, or applying formulas. |
| Error | Ingestion Run Failed | A critical failure occurred during file transfer, schema validation, or parsing; requires immediate remediation. |
| Paused | Temporarily Suspended | Administrator has paused the schedule (e.g., during upstream system maintenance) without deleting configuration. |
Drilling into Refresh History and Run Logs
To diagnose stream health, administrators navigate to the Refresh History tab within any Data Stream record. The Refresh History table provides an immutable audit trail of every ingestion execution:
- Start Time & End Time: Execution duration and processing timestamps (in UTC).
- Status: Visual indicator showing Success, Partial Success, or Failed.
- Records Processed: Total count of source records successfully ingested and committed to the underlying Data Lake Object (DLO).
- Records Failed: Count of records rejected by the ingestion engine due to validation or parsing errors.
- Run Message / Error Description: Technical diagnostic details explaining failure causes (e.g.,
Invalid timestamp format in row 412,Schema validation failed: Column 'EMAIL_ADDR' missing from source header). - Error Log / Manifest Download: For batch file ingestion (such as S3 or SFTP), Data Cloud generates a downloadable CSV error manifest detailing the exact line numbers, source records, and specific validation failure reasons.
Inspecting Data with Data Explorer
Data Explorer is a high-performance administrative inspection utility built directly into Data Cloud. It enables technical users to query and inspect raw lakehouse records without requiring external BI tools or SQL workbench connections.
Access Governance & Supported Object Types
As established in access governance, Data Explorer is restricted to the Data Cloud Architect permission set (and the Legacy Data Aware Specialist set in orgs that have not migrated). Activation and consumption roles do not have access to this tool.
Data Explorer supports querying three distinct layers of the Data Cloud architecture:
- Data Lake Objects (DLOs): The raw, untransformed staging tables populated directly by data streams.
- Data Model Objects (DMOs): The harmonized canonical Customer 360 entities (standard and custom) populated by mapping definitions.
- Calculated Insights (CIs): The multi-dimensional metrics and aggregations generated by analytical queries.
Operational Features of Data Explorer
- Sample Row Inspection: Displays up to 100 sample records per query, optimized for rapid schema validation and data hygiene checks.
- Column Filtering & Ordering: Allows users to select specific fields, reorder columns, and apply filter criteria (e.g.,
WHERE CreatedDate > 2026-09-01orWHERE Customer_Status__c = 'Active'). - Data Type & Format Verification: Enables consultants to verify that date fields parsed as Datetime rather than Text, that numeric currency values contain no unexpected string symbols, and that formula fields evaluated correctly.
- Mapping Validation: By inspecting a DLO side-by-side with its target DMO in separate browser tabs, consultants can verify whether source records are correctly populating target canonical fields or whether unmapped fields are resulting in null values.
Inspecting Customer 360 with Profile Explorer
While Data Explorer inspects tabular lakehouse records, Profile Explorer is purpose-built to inspect consolidated Customer 360 Unified Profiles. It provides a 360-degree interactive view of individuals synthesized by Identity Resolution rulesets.
User Access & Search Paradigms
Profile Explorer is broadly accessible across technical and business roles, including Data Cloud Architect and Data Cloud User, with the activation roles able to view unified profiles they are building audiences against.
Users search for unified profiles using specific search keys:
- Unified Individual ID: The unique system identifier generated by Data Cloud's identity resolution engine.
- Contact Points: Email address, phone number, or physical street address.
- Party Identification: External system identifiers such as CRM Contact ID, Loyalty Account Number, or National ID.
Core Profile Explorer Views & Diagnostic Functions
When a unified profile is opened, Profile Explorer renders three foundational panes:
- Unified Attributes Card: Displays the single reconciled view of customer demographic fields (e.g., Unified First Name, Unified Last Name, Unified Household Income). Consultants use this view to verify that Reconciliation Rules (Most Frequent, Source Priority, or Last Updated) are selecting the correct golden values from contributing source records.
- Unified Contact Points: Lists all email addresses, phone numbers, and physical addresses linked to this unified individual across all connected source systems. Crucial for verifying that contact point consolidation rules are functioning properly.
- Engagement Timeline: Chronologically visualizes every interaction and transactional event associated with this individual (e.g., website clicks, mobile app telemetry, closed-won opportunities, service case interactions). Consultants verify that engagement DMOs are correctly linked to the Unified Individual via foreign key relationships.
Diagnosing Identity Resolution Quality: Over-Grouping vs. Under-Grouping
Profile Explorer is the primary diagnostic tool for identifying identity resolution configuration flaws:
- Over-Grouping (False Merges): Occurs when overly permissive match rules (such as matching on fuzzy name alone or shared corporate phone numbers) mistakenly merge two distinct people into a single Unified Individual. In Profile Explorer, this is revealed when an individual's card displays conflicting names or disparate email domains.
- Under-Grouping (Missed Merges): Occurs when overly strict match rules fail to link duplicate records representing the same person. In Profile Explorer, searching for a customer's known email returns multiple separate Unified Individual IDs instead of a single consolidated profile.
Comparison Matrix: Data Explorer vs. Profile Explorer
The following matrix highlights the critical distinctions tested on the certification exam:
| Architectural Dimension | Data Explorer | Profile Explorer |
|---|---|---|
| Target Data Layer | Raw DLOs, Canonical DMOs, Calculated Insights | Unified Individual DMO, Unified Contact Points, Unified Engagement |
| Primary Objective | Data pipeline debugging, mapping verification, raw schema validation | Customer 360 inspection, identity resolution verification, customer service lookup |
| Data Structure | Tabular, relational row-and-column grid (up to 100 sample rows) | Consolidated entity card with interactive tabs and engagement timeline |
| Search Paradigm | Object selection with SQL-like field filters | Direct search by Contact Point (Email, Phone) or Party Identifier |
| Authorized Roles | Data Cloud Architect only (plus Legacy Data Aware Specialist) | Data Cloud Architect, Data Cloud User, activation roles in view mode |
| Identity Context | Displays individual source records prior to unification | Displays unified golden record after ruleset consolidation |
Usage Metrics, Storage Tiers & Data Cloud Credit (DCC) Consumption
Salesforce Data Cloud operates on a consumption-based pricing model metered through Data Cloud Credits (DCC). System administrators must proactively monitor DCC burn rates to optimize performance, prevent budget overruns, and maintain compliance with licensing contracts.
The Four Primary Consumption Engines
Data Cloud Credits are consumed across four core functional operations:
- Data Ingestion:
- Streaming Ingestion: Metered per million streaming events ingested via the Web & Mobile SDKs or Ingestion API.
- Batch File Ingestion: Metered based on data volumes (in gigabytes) processed from Amazon S3, Google Cloud Storage, Microsoft Azure Blob, or external SFTP servers.
- Salesforce CRM Connectors: Standard change data capture (CDC) ingestion from connected Salesforce orgs draws from standard platform entitlements.
- Data Processing & Transformations:
- Batch Data Transforms: Executing Apache Spark-powered data recipes and complex multi-join data transformations consumes compute credits based on execution runtime.
- Calculated Insights (CIs): Processing multidimensional metrics across billions of rows consumes processing credits. Insights scheduled to refresh every hour consume significantly more credits than those scheduled daily.
- Identity Resolution:
- Metered based on the total volume of source records evaluated and consolidated across Identity Resolution ruleset processing cycles.
- Segmentation & Activation:
- Segment Processing: Metered based on segment refresh frequency (e.g., standard 12/24-hour schedules vs. rapid 1-hour/4-hour refreshes) and the complexity of filtering criteria.
- Activation Publishing: Generating and deploying filtered audience payloads and attributes to external activation targets.
Storage Tiers & Governance Tools
- Active Lakehouse Storage: High-performance Parquet format storage within Hyperforce object storage, supporting rapid querying, segmentation, and insight calculations.
- Salesforce Digital Wallet: A centralized administrative console within Salesforce Setup that provides near real-time visibility into Data Cloud Credit consumption, consumption trends, and burn rate forecasts across billing cycles.
Troubleshooting Common Ingestion & Processing Errors
When data streams enter an Error state or records are rejected into error logs, consultants must quickly pinpoint the failure mechanism. The following reference table outlines the most frequent ingestion errors encountered in production and on the certification exam:
| Failure Category | Root Cause & Technical Mechanism | Diagnostic Indicator in Run Log | Remediation Procedure |
|---|---|---|---|
| Datetime Parsing Failure | Source timestamp does not conform to ISO 8601 UTC format (e.g., MM/DD/YYYY hh:mm A instead of YYYY-MM-DDTHH:mm:ss.SSSZ). | Error: Unable to parse field 'Transaction_Date' as Datetime | Author a Formula Field on the stream using date parsing functions (e.g., PARSE_DATE) to convert string into standard ISO format before mapping. |
| Schema Mismatch (Missing Columns) | The header in a newly ingested CSV or Parquet file is missing a column defined in the Data Stream schema. | Error: Column 'CUSTOMER_ID' defined in schema not found in source file | Update upstream extraction file export to restore missing column, or edit Data Stream schema to remove or mark field optional. |
| Schema Mismatch (Extra Columns) | Source file contains unexpected columns not declared during initial data stream schema setup. | Warning: Unrecognized columns ignored or Error: Schema validation mismatch | Ingest additional fields by editing the Data Stream schema, adding the new source attributes, and saving the updated definition. |
| Primary Key Collision | A batch upload contains multiple distinct records sharing the exact same primary key value within a single file. | Warning: Duplicate primary key detected. Last record retained. | Cleanse source extract to ensure true uniqueness, or create a composite primary key in the stream by concatenating multiple fields (e.g., AccountId + '_' + SystemId). |
| Numeric / Currency Cast Error | A numeric or currency field contains alphabetic characters, currency symbols ($, €), or thousand separators (,). | Error: Cannot cast string '$1,250.00' to Decimal | Use a string manipulation formula field (REPLACE, TRIM) to strip non-numeric characters before mapping to a numeric canonical field. |
| Null Pointer in Transform | A formula field or SQL batch transform evaluates a function against a null attribute without handling null safety. | Error: NullPointerException during formula evaluation on field 'Full_Name' | Implement conditional logic (IF(ISNULL(source_field), '', source_field)) to ensure formulas gracefully handle null values. |
A daily batch data stream ingesting customer transaction files from an Amazon S3 bucket enters an 'Error' status. An administrator reviews the Refresh History log and observes that 50,000 records were rejected with the message: 'Unable to parse field Transaction_Timestamp as Datetime'. Inspection reveals the S3 CSV files supply timestamps formatted as '10/24/2026 03:45 PM'. What is the root cause of this failure, and how should it be resolved?
A customer service director reports that a customer's profile card in Service Cloud displays an outdated mobile phone number and contradictory home addresses, despite recent profile updates across two external systems. A Data Cloud consultant needs to investigate whether Identity Resolution correctly merged the customer's records and verify which source systems contributed the phone number and addresses. Which tool should the consultant use, and what information will it provide?
An enterprise organization with hundreds of millions of engagement records observes an unexpectedly rapid consumption of Data Cloud Credits (DCC) over a two-week period. An architectural review identifies several complex Calculated Insights (CIs) that aggregate customer engagement scores across three years of historical data. What architectural change will most effectively reduce DCC processing consumption without eliminating the insights?