Troubleshooting practice questions

Reconciliation rules were set to Most Recent instead of Source Sequence.

Why it fails: Reconciliation rules only decide which source value survives once records already match; they never determine whether a match occurs. Source Sequence versus Most Recent affects field survivorship, so this cannot explain records with identical emails failing to merge.

Data Cloud requires manual SQL triggers to activate e-commerce data streams.

Why it fails: Data Cloud ingests and harmonises streams through configured connectors and data streams; no manual SQL triggers exist for activation. This is tempting because stream activation sounds relevant, but identity resolution merges records via match rules, not SQL execution.

The Unified Individual DMO is missing an activation target assignment.

Why it fails: Activation target assignment governs where a Unified Individual is published, such as to a data space or activation platform. It has no bearing on match processing; failed merges despite matching emails point to match rule or reconciliation configuration instead.

Increase data stream ingestion frequency from hourly to real-time.

Why it fails: Real-time ingestion increases processing volume and load, worsening the memory and timeout exception. It is tempting because real-time streams suit low-latency use cases, and would be correct when the business requires immediate data availability rather than batch processing.

Restart the core Salesforce CRM production server.

Why it fails: Restarting the CRM server does not address the Identity Resolution batch job's memory or timeout limits, which run on Data 360 infrastructure. It is tempting because restarting clears transient faults, and would be correct if the CRM application itself were unresponsive.

Switch the reconciliation rule to Highest Frequency.

Why it fails: Highest Frequency changes which value survives reconciliation, not the match rule's memory footprint, so the timeout persists. It is tempting because reconciliation settings affect job behaviour, and would be correct when resolving conflicting field values between matched profiles.

Check the core CRM Apex code coverage for all triggers in the organization.

Why it fails: Apex code coverage has no impact on Data Cloud CRM data streams.

Verify that all Data Model Objects have been converted to custom Salesforce objects in Object Manager.

Why it fails: Data Model Objects are native to Data Cloud and are not converted to core CRM custom objects.

Calculated Insights stripped out historical records older than 90 days.

Why it fails: Calculated Insights compute derived metrics; they do not delete source records or sever foreign keys to Unified Individual. It is tempting because retention windows exist elsewhere in Data 360, but the missing keys stem from identity resolution failing to match profiles, not from insight retention.

Batch segmentation only processes the first 10 million records by default.

Why it fails: Batch segmentation has no fixed 10-million-record cap; it processes the full segment population. It is tempting because large volumes do affect processing time, but the missing foreign keys indicate identity resolution did not link those Engagement DMO records to a Unified Individual, not a row limit.

Streaming segments overwrite batch segment membership automatically.

Why it fails: Streaming and batch segments maintain separate membership; streaming does not overwrite batch results. It is tempting because both segment types can target the same population, but the absent foreign keys point to identity resolution failing to unify those profiles, not segment-type interference.

Recreate the Identity Resolution ruleset to clear transaction locks.

Why it fails: Identity Resolution rulesets govern profile consolidation, not outbound HTTP connectivity, so recreating them cannot fix a webhook timeout. It is tempting because ruleset rebuilds clear stuck transactions, and would be correct where activation failures stem from unresolved identity conflicts rather than network reachability.

Increase the Data Cloud activation target chunk size to 1 million records per request.

Why it fails: Raising chunk size to a million records enlarges each payload, worsening the timeout by increasing request duration and memory pressure. Large chunks suit high-throughput batch targets, and would be correct where the endpoint tolerates long requests and the bottleneck is per-request overhead, not latency.

Switch the activation target type from webhook to cloud storage.

Why it fails: Switching to cloud storage abandons the REST webhook entirely, so the timeout against the custom endpoint is never diagnosed or resolved. Cloud storage targets suit bulk file delivery, and would be correct if the requirement were batch export rather than real-time webhook delivery.

Switch all reconciliation rules from Source Sequence to Longest String.

Why it fails: Longest String selects the longest value per field, which does not address duplicate profiles arising from weak match rules or missing normalisation; it may even merge or retain inconsistent values. Source Sequence is correct when a defined source priority must determine the winning attribute value.

Delete all data spaces and restart the entire Salesforce CRM instance.

Why it fails: Deleting data spaces and restarting the CRM destroys the ingested data and configuration that troubleshooting requires, and cannot reduce inflated counts caused by reconciliation or matching rules. It is tempting as a reset when a sandbox is corrupted, but production identity resolution demands diagnostic analysis, not teardown.

Verify that the Activation Target authorization token is active.

Why it fails: Activation Target tokens govern outbound data delivery to external platforms, which occurs after a Calculated Insight has already computed. A wrong or stale result points to the insight's definition, filters or source data, not to downstream activation credentials. Token validity is the right check when an activation fails to deliver data.

Check that the underlying Data Stream ingestion schedule is set to streaming mode.

Why it fails: Data Streams ingest raw source data; forcing streaming mode does not correct a Calculated Insight's logic, filters or aggregation. The insight's own definition and its source object's data freshness are what determine output. Streaming mode is the right setting when near-real-time ingestion of event data is genuinely required.

Confirm that the Segment Canvas refresh schedule matches the Calculated Insight run time.

Why it fails: Calculated Insight results depend on its own definition, filters and source data freshness, not on when a Segment Canvas refreshes; the Canvas consumes segments rather than feeding the insight. Schedule alignment is relevant when troubleshooting a segment's membership appearing stale in an activation, not the insight's computed output.

The Einstein Studio model training metrics.

Why it fails: Einstein Studio training metrics describe model training runs, not the data pipeline stages an activation traverses while Publishing, so they expose no bottleneck. It tempts because Einstein Studio is where Data 360 model behaviour is inspected, and would be correct if the delayed activation stemmed from a failing or slow model training job.

The Salesforce Core Setup Audit Trail.

Why it fails: The Setup Audit Trail records configuration changes made by administrators, not activation pipeline execution stages, so it cannot reveal where a Publishing job is stalled. It tempts because it is the standard Salesforce place to trace who changed what, and would be correct if the issue were an unexpected metadata or permission change.

The Marketing Cloud Journey Builder analytics dashboard.

Why it fails: Journey Builder analytics reports marketing journey performance for Marketing Cloud, not Data 360 activation pipeline execution, so it shows no Publishing bottleneck. It tempts because activation output often feeds journeys, and it would be correct if the problem were journey entry or send throughput rather than the activation itself.

Marketing Cloud Engagement requires Real-Time Data Graph activation for all Calculated Insights.

Why it fails: Real-Time Data Graph activation is not a prerequisite for Calculated Insights. The real cause is that serialised custom attributes must be activated and mapped to the Marketing Cloud Engagement Data Extension; without that mapping, attributes never appear. Data Graph activation addresses a different capability.

Calculated Insights cannot be exported to Marketing Cloud Engagement due to API limits.

Why it fails: Calculated Insights are exportable to Marketing Cloud Engagement; no such API limit exists. The genuine cause is that serialised custom attributes require the Calculated Insight to be activated through a Data Cloud data action or activation, with the attributes mapped into the target Data Extension. API limits are not the mechanism.

Activation targets only support streaming data streams, not batch-based Calculated Insights.

Why it fails: Activation targets accept both batch and streaming sources, including Calculated Insights. The actual failure is that serialised custom attributes need explicit mapping and activation configuration before they populate Marketing Cloud Engagement Data Extensions. Streaming-only support is not a real restriction.

Restart the Kafka message broker cluster via the Data Cloud setup menu.

Why it fails: Data Cloud ingests from cloud storage buckets directly; no Kafka broker sits in that path, so restarting one cannot clear a bucket connectivity or parsing fault. It is tempting because broker restarts do resolve stalled Kafka-based streaming connectors, which would be the right move only when the source is an actual Kafka topic.

Rebuild the Identity Resolution ruleset associated with the data stream.

Why it fails: Identity Resolution consolidates matched profiles after ingestion; it does not govern stream connectivity, credentials or schema parsing, so rebuilding the ruleset leaves the failing stream unchanged. It is tempting because ruleset errors do surface as downstream data quality failures, making it the correct fix when ingestion succeeds but profiles merge incorrectly.

Check the Apex Debug Logs in Salesforce Setup.

Why it fails: Apex Debug Logs capture Salesforce application execution, not external cloud storage ingestion, so they reveal nothing about the failing data stream. It is tempting because debug logs are the standard diagnostic for Apex errors, but Data Cloud connector failures surface in Data Cloud ingestion job monitoring instead.

Configuring an activation target with an invalid webhook URL.

Why it fails: Calculated Insight failures stem from SQL errors such as invalid object references, unsupported functions or missing fields, not from activation webhook configuration. It is tempting because a malformed webhook URL does break activation, and would be correct if the question asked which factor causes activation or data-action failures rather than SQL execution errors.

Failing to enable Salesforce Shield Platform Encryption on the user profile.

Why it fails: Platform Encryption protects data at rest and does not affect whether Calculated Insight SQL compiles or executes; failures arise from query and schema errors. It is tempting because encryption settings can restrict field access, and would be correct if the question asked why a user cannot view encrypted fields rather than why SQL execution fails.

The segment was published during off-peak hours.

Why it fails: Publishing time affects processing load and latency only, not row cardinality in the export. It is tempting because off-peak scheduling is a genuine tuning practise for large batch segments, and would be the right answer if the question asked how to reduce activation processing contention.

The Salesforce CRM license count is too low.

Why it fails: Licence counts govern user provisioning, not export row generation, so they cannot duplicate segment members. It is tempting because licence limits do block activation in some platforms, and raising a seat count would be the right fix when an activation fails outright for insufficient Salesforce licences.

The segment type is set to streaming instead of batch.

Why it fails: Streaming versus batch changes how often a segment recalculates, not whether a profile appears once per row; duplicates stem from the identity resolution or activation target key. Streaming would be correct where near-real-time segment membership is required for time-sensitive journeys.

Restart the entire Data Cloud tenant instance from the Setup menu.

Why it fails: Tenant-wide restarts disrupt every stream and user without diagnosing the failed one; troubleshooting requires inspecting that stream's run history, error logs, and source connectivity. It tempts as a quick reset, but a full instance restart is a platform-wide action, not a targeted diagnostic step.

Manually rewrite the underlying Apache Spark cluster configuration files.

Why it fails: Spark cluster configuration is platform-managed in Data Cloud; consultants cannot edit underlying files, and doing so would not address a stream's source, schema, or mapping failure. It tempts because Spark powers ingestion, but the actionable fix lies in stream configuration and error logs.

Setup > Apex Classes > IngestionAPIController > Schema

Why it fails: Apex Classes hold executable Apex code, not the declarative schema contract enforced by the Ingestion API, so no IngestionAPIController class exposes the expected payload definition. It tempts because Salesforce administrators often inspect Apex to trace integration behaviour, making a controller class a superficially reasonable place to look for schema logic.

Data Cloud Setup > Data Spaces > API Management > Schema Explorer

Why it fails: Data Cloud Setup exposes Ingestion API schema definitions under Data Spaces, but the Schema Explorer path shown here is not where the connector's expected payload schema is published. It tempts because Data Spaces configuration is where ingestion sources are registered, so administrators reasonably expect schema metadata to live alongside them.

Setup > Integration > API Schemas > Salesforce Data Cloud

Why it fails: Setup > Integration > API Schemas > Salesforce Data Cloud does not exist as a schema inspection location; Ingestion API schemas are surfaced through Data Cloud's own setup area rather than a generic Integration node. It tempts because Integration menus commonly host API artefacts, so the naming feels plausible to administrators familiar with Salesforce setup navigation.

The Connected App callback URL and OAuth scopes

Why it fails: The callback URL and OAuth scopes govern inbound user authentication to Salesforce, not the outbound credentials Data Cloud presents to S3. They are tempting because authentication errors point at OAuth, and they would be correct when a connected app's own login flow fails.

The Marketing Cloud Connector user profile settings

Why it fails: The Marketing Cloud Connector authenticates Marketing Cloud journeys, not S3 exports; the activation's S3 access uses an AWS credential configured on the activation target. It is tempting because connector user profiles govern cross-cloud authentication, and would be correct if the failing channel were a Marketing Cloud journey rather than a direct S3 file drop.

The Einstein Trust Layer activation guardrails

Why it fails: Einstein Trust Layer guardrails govern model prompting, toxicity and data masking, not S3 credential validation, so they cannot cause or explain an authentication failure. They are tempting because they are a Data Cloud governance feature, and would be relevant when auditing generative AI output rather than an activation export.

Review the Apex Job Queue for batch processing failures.

Why it fails: Apex Job Queue holds Salesforce platform batch jobs, not Data Cloud activation runs. It is the right place to inspect Apex batch failures, such as governor-limit exceptions, but activation export errors surface in Data Cloud activation logs and S3 target diagnostics.

Examine the Streaming Ingestion monitor in Data Cloud.

Why it fails: The Streaming Ingestion monitor tracks real-time ingestion into Data Cloud, not outbound batch exports to Amazon S3, so it holds no export-failure detail. It is tempting because it surfaces ingestion errors, and would be the right place when diagnosing streaming data failing to land in Data Cloud itself.

Check the Setup Audit Trail in the core Salesforce CRM org.

Why it fails: The Setup Audit Trail records configuration changes in the core CRM org, not Data Cloud activation job execution. Export failures and skipped records are diagnosed in Data Cloud's activation job history and error logs. Tempting because it is a familiar troubleshooting location, but it holds no activation run details.

Unix Epoch timestamp integer (e.g., 1698000000)

Why it fails: Unix epoch integers carry no timezone or ISO 8601 structure, so Data Cloud's timestamp validation rejects them. It is tempting because epoch values are compact and common in application logs and APIs, and they would be correct after conversion into an ISO 8601 string with offset.

RFC 2822 email date format

Why it fails: RFC 2822 is an email header date syntax, not the ISO 8601 profile Data Cloud validates against, so payloads are rejected. It is tempting because it is a recognised internet date standard appearing in message metadata, and it would be correct for parsing email headers rather than Data Cloud ingestion timestamps.

US Standard format (MM/DD/YYYY HH:MM AM/PM)

Why it fails: Data Cloud requires ISO 8601 timestamps; US-style MM/DD/YYYY with AM/PM is ambiguous and fails schema validation. It is tempting because it is the common convention in North American source systems and reports, and it would be correct only where a transformation step converts it to ISO 8601 before ingestion.

It schedules the batch frequency of data stream refreshes from source systems.

Why it fails: Activation membership determines which profiles and attributes are published to an activation target; it does not govern ingestion timing. Refresh cadence for data streams is configured on the data stream itself, so this option mistakes outbound activation scoping for source-side scheduling.

It defines the structural schema of the Data Model Object used for ingestion.

Why it fails: An activation membership selects which unified profiles and related attributes flow to an activation target, governing downstream segmentation output. Defining a Data Model Object's structural schema is the job of the DMO itself during data mapping, so this option confuses outbound activation targeting with ingestion-time modelling.

It acts as the encryption key for secure cloud storage activation targets.

Why it fails: Activation membership defines which unified profile records are included in an activation target's audience; it is not a cryptographic key. It tempts because activation targets deliver data externally, where encryption matters, but key management is handled separately, not by the membership itself.

Check the Salesforce Core Setup Audit Trail for login failure events.

Why it fails: The Setup Audit Trail records configuration and administrative changes, plus login events, not activation publish record failures. It is the right place to trace who altered metadata or when a user signed in, but it holds no per-record export error detail for the activation job.

Open the Marketing Cloud Engagement Tracking Report for email sends.

Why it fails: Marketing Cloud Engagement tracking reports cover email sends, opens and clicks, not Data 360 activation export failures. They would be correct for diagnosing deliverability or engagement metrics, but they contain no information about which records failed to publish from an activation.

Examine the Data Stream Ingestion status dashboard for failed CSV files.

Why it fails: Data Stream Ingestion status covers inbound data loads into Data 360, not outbound activation publishing. It is the correct place to diagnose failed CSV ingestion into a data stream, whereas the publish job's partial failures require the activation's own error logs or activation run details.

Increase the activation target polling frequency.

Why it fails: Zero unified profiles points to mismatched match rules, missing primary keys or unmapped source objects; activation polling affects downstream segment publishing, not profile creation. It tempts because polling frequency is a tunable setting, but it cannot reconcile identities that never unified.

Delete all data spaces and reinstall the Data Cloud package.

Why it fails: Deleting data spaces and reinstalling the package destroys configuration and does not address match-rule or data-stream mapping faults causing zero profiles. It tempts as a reset, but identity resolution failures are diagnosed by inspecting rulesets, source priorities and ingestion, not by reinstallation.