MB-260 flashcards
MB-260 · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- Customer data platform
- A system that ingests data from multiple sources, unifies records into customer profiles, and activates those profiles to downstream systems. Customer Insights - Data is Microsoft's CDP.
- Unification process
- The pipeline step that matches records across source tables using match rules and merges their field values into a single unified customer profile.
- Environment capacity
- The provisioned limit on unified profiles and processing for a Customer Insights environment. Exceeding it causes refresh or processing failures until capacity is increased.
- Data residency region
- The geographic region chosen when the environment is provisioned. It determines where data is stored and cannot be changed after provisioning.
- Administrator role
- The Customer Insights role that manages data sources, connections, environment settings, and user access. Contributors build segments and measures; Viewers have read-only access.
- Activation connector
- A configured connection that delivers segments or profiles to an external system such as Dynamics 365, advertising platforms, or a data lake.
- Data source vs table
- A data source is the container for a connected system; each table inside it maps to one source entity and requires a primary key.
- Power Query ingestion
- The connector-based ingestion path that lets you transform, rename, and filter columns at load time before data lands in the unified store.
- Incremental refresh
- A refresh mode that loads only rows changed since the last run, using a watermark or modified-date column. It is cheaper than a full reload.
- Data profiling
- Pre-unification analysis showing null percentages, distinct counts, and type mismatches per column, so schema problems are fixed at the source table.
- Schema drift
- A change in the source structure, such as a renamed or removed column, that breaks the mapping and causes ingestion or refresh failures.
- Duplicate primary keys
- A common ingestion failure where two rows share the same key value in a table. Each row must have a unique primary key.
- CSV upload limits
- Manual file uploads are capped by size and row count. Larger volumes should use a connector or the ingestion API instead of manual upload.
- Match rule
- A unification rule that pairs a source table with a match column and normalization method to decide which records represent the same person.
- Merge rule
- A unification rule that decides which value wins when matched records disagree on a field, for example by source priority or most recent value.
- CustomerId
- The system-generated identifier assigned to each unified profile. It is the key used by segments, measures, and exports to reference a profile.
- Fuzzy matching
- A match method that tolerates typos and formatting differences. It raises match rates but increases false positives compared with exact matching on normalized keys.
- Unification report
- A report showing matched versus unmatched record counts, match rate per rule, and duplicate clusters, used to tune thresholds before go-live.
- Rerun after rule change
- Changing match or merge rules does not retroactively alter existing profiles. Unification must be rerun before the new rules take effect.
- Activity timeline
- A chronological view of a profile's interactions, built from activity tables linked to the profile through a shared key column.
- Custom enrichment
- Bring-your-own enrichment that joins external data to unified profiles on a key such as email or address, adding fields for segmentation.
- Microsoft enrichment
- Optional, cost-bearing enrichment that appends Microsoft brand and interest data to unified profiles, optionally scoped to a filtered subset.
- Rule-based segment
- A segment built in the visual builder using conditions on unified fields and measures, with AND/OR groups and include or exclude set operations.
- Code-based segment
- A segment defined by a SQL-like query over the unified profile entity. The query must return the CustomerId column and is validated before activation.
- Segment refresh
- The run that recomputes segment membership. Membership is a snapshot as of the last successful refresh and does not update continuously.
- Segment activation requirement
- A segment cannot be delivered to a destination until an export or connection is configured. Without one, the segment exists but stays internal.
- Segment nesting
- Grouping conditions inside other conditions in the segment builder. Nesting is allowed, but deep nesting degrades segment evaluation performance.
- Measure definition
- A calculation over unified profiles and related tables, specified by name, source entity, aggregation such as sum or average, and optional filter or group-by.
- Prediction model
- A machine learning model trained on historical unified data to score outcomes such as churn or lifetime value, writing results back as profile fields.
- Prediction training set
- The labeled historical data used to train a prediction model. It must include the target outcome and sufficient examples for the model to learn.
- Export definition
- The configuration that specifies which entity and fields to push to a destination such as Azure Blob or Dataverse, plus the refresh schedule.
- Export run history
- A log of past export runs including status and error detail. Failed exports retain their error messages here for troubleshooting.