แก้ไข

แชร์ผ่าน

Troubleshoot unification output tables in Dynamics 365 Customer Insights - Data

  • บทความ

Introduction to output tables

Each step of the data unification process produces system-generated output tables.

To troubleshoot an unexpected unification result, you can trace these output tables:

Step Table Description
Deduplication Deduplication_{Datasource}_{Table} Deduplicated records for each source table
Match ConflationMatchPairs Matched records across source tables
Merge Customer Unified customer profiles

For detailed descriptions of each output table, see Understand output tables.

View output tables

Method Size Deduplication ConflationMatchPairs Customer
Download 100K records Top 100K ✔ ✔ ✔
Set up exports Full ✔ ✔ ✔
Create a sandbox environment Full ✔ ✔ ✔

Method 1: Download 100K records

Use this method if there are less than 100,000 records.

In Customer Insights - Data, open the Tables page. For each output table, select Download to get the most recent 100,000 records of the table.

Screenshot that shows how to download records of an output table.

To make sure this method contains all output information, see Method 3: Create a sandbox environment.

Method 2: Set up exports

Use this method if there are more than 100,000 records.

For each output table, set up an export. We recommend exporting to blob storage.

Refresh all exports, and then the full tables are written to the configured location.

Method 3: Create a sandbox environment

Create a new sandbox environment to re-create the unification configuration on tables that contain a subset of problem records of the original tables.

This step ensures that Method 1: Download 100K records contains all output information.

Data sources

Create new tables that contain only a small subset of the problem records. We recommend less than 100,000 records.

Refresh the tables to ingest them into Customer Insights - Data.

Unify the problem tables

  1. Map the problem tables.
  2. Copy the original deduplication rules to the problem tables.
  3. For each problem table:
    • Copy the original matching rules.
    • Enable Include all records for better visibility into the data.
  4. In the Unified data view step, keep the default configuration.
  5. Run unification to generate the customer profiles.

Resolve unification results

Depending on the location of your unexpected result, you might need to verify different output tables.

See Example of explaining a unification result.

Deduplication

To verify if deduplication is behaving as expected, check:

  • Source data of the problematic result
  • Deduplication configuration

Make sure to consider all configurations, such as:

  • Normalization
  • Precision
  • Exceptions
  • Merge preferences

See an overview of deduplication concepts in Define deduplication rules and the example in Deduplication concepts and scenarios.

Match

To verify if match is behaving as expected, check:

  • Source data of the problematic result
  • Related Deduplication records
  • Match configuration

Make sure to consider all configurations, such as:

  • Match order
  • Enrichments
  • Normalization
  • Precision
  • Exceptions
  • Custom match conditions
  • Merge preferences

See an overview of match concepts in Define matching rules for data unification.

Merge

To verify if merge is behaving as expected, check:

  • Source data of the problematic result
  • Related Deduplication records
  • Related ConflationMatchPairs records
  • Merge configuration

Make sure to consider all configurations, such as:

  • Excluded fields
  • Clusters
  • Merge preferences
  • Grouped fields
  • Custom ID generation.

See an overview of the merge behavior in Unify customer columns for data unification and the example.

Understand output tables

Deduplication tables

The Deduplication table is the source table deduplicated by the configured rules. If no configured rules exist, the source table is deduplicated on the columns referenced in the match rules.

Column Source Type Description
PrimaryKey Source String The configured source primary key
PrimaryKey_Alternate System String The concatenated list of primary keys identified for a deduplication group
DeduplicationGroup ... DeduplicationGroup_N System String The identifier for the group of similar records based on deduplication rules
Rule ... Rule_N System String The deduplication rule that the deduplication group matched
Score ... Score_N System Double The score returned for the deduplication rule
Deduplication_WinnerId System String The winning primary key for the deduplication group
Other mapped fields Source Various The remaining mapped fields from the source table

ConflationMatchPairs table

The ConflationMatchPairs table is the set of matched deduplicated records based on the configured rules.

Column Source Type Description
TrueObjectId System String The temporary identifier for records matched across source tables
PrimaryKey ... PrimaryKey_N Source String The matched source primary keys
PrimaryKey_Alternate ... PrimaryKey_Alternate_N System String The alternate keys for the matched source primary keys
ConflationMatchPairs_ModifiedOn System Date & time The timestamp of the most recent change to this matched record
Other matched fields Source Various The remaining mapped fields from the source tables

Customer table

The Customer table is the final set of customer profiles produced by merging the source columns from ConflationMatchPairs based on the configuration of the unified fields.

Column Source Type Description
CustomerId System String The unique guid identifier of the profile
PrimaryKey ... PrimaryKey_N Source String The matched source primary keys
PrimaryKey_Alternate ... PrimaryKey_Alternate_N System String The alternate keys for the matched source primary keys
Unified fields Source Various The final fields determined by applying the unified field configuration to the source fields

Example of a unification result

Source tables

MyData_Contact

ContactId FirstName LastName Email
1 Thomson monica.thomson@contoso.com
2 Monica Smith monica.thomson@contoso.com

MyData_Referral

Id FirstName LastName EmailAddress ReferralDate
100 Moni Thomson monica.thomson@contoso.com January 1, 2024 12:00 AM
200 Monica Smith monica.thomson@contoso.com December 24, 2020 12:00 AM

Deduplication tables

Deduplication_MyData_Contact

If you deduplicate on Email:

ContactId ContactId_Alternate Deduplication_GroupId Rule Score Deduplication_WinnerId FirstName LastName Email
1 1;2 guid() DedupOnEmail 1.0 1 Thomson monica.thomson@contoso.com

Deduplication_MyData_Referral

If you deduplicate on EmailAddress:

Id Id_Alternate Deduplication_GroupId Rule Score Deduplication_WinnerId FirstName LastName EmailAddress ReferralDate
100 100;200 guid() DedupOnEmailAddress 1.0 100 Moni Thomson monica.thomson@contoso.com January 1, 2024 12:00 AM

ConflationMatchPairs table

If you match on Contact.Email == Referral.Email:

TrueObjectId Contact_ContactId Contact_ContactId_Alternate Referral_Id Referral_Id_Alternate ConflationMatchPairs_ModifiedOn Contact_FirstName Contact_LastName Contact_Email Referral_FirstName Referral_LastName Referral_EmailAddress Referral_ReferralDate
1__00 1 1;2 100 100;200 now() Thomson monica.thomson@contoso.com Moni Thomson monica.thomson@contoso.com January 1, 2024 12:00 AM

Customer table

If you set up the unified fields as follows:

  • FirstName: Contact.FirstName is prioritized over Referral.FirstName.
  • LastName: Contact.LastName is prioritized over Referral.LastName.
  • Email: Contact.Email is prioritized over Referral.EmailAddress.
  • ReferralDate: Referral.ReferralDate is taken.
CustomerId Contact_ContactId Contact_ContactId_Alternate Referral_Id Referral_Id_Alternate FirstName LastName Email ReferralDate
guid() 1 1;2 100 100;200 Moni Thomson monica.thomson@contoso.com January 1, 2024 12:00 AM