AI-powered customer identity management.
AI-powered identity management that correctly identifies your customers.
Learn more about how Stitch correctly identifies your customers.
Apply semantic tags to your data in feeds or custom domain tables.
Assign a unique ID to each unique individual that is discovered in your data.
Review every connection. See how customer identities were resolved.
Apply to all fields that contain PII.
A feed will infer tags for fields that contain PII.
Reshape data with Spark SQL, then apply tags fields that contain PII.
What should you look at during Stitch QA?
It’s important to look for anomalies in Stitch output.
A blocking key defines combinations of characters for blocking strategies.
Customer keys represent unique IDs in your source data.
Foreign keys act as primary keys and are used for deterministic matching.
False positives occur when records are incorrectly added to clusters.
Given name variations can prevent accurate clustering.
Low-scoring record pairs should be investigated to validate accuracy.
Consistent application of tags across sources builds better customer profiles.
False negatives that occur when distinct records are incorrectly split.
Identify values to add to your bad-values and 360 values blocklists.
Exclude values that do not improve the quality of your customer profiles.
Include values for ID resolution; exclude them from your Customer 360.
Variations in given names are associated with the same individual.
Blocking strategies determine how records are linked during ID resolution.
Identify foreign keys that contain case-sensitive values.
Empty tables can be present with CCPA and GDPR workflows.
Email address with certain patterns will be ignored by Stitch.
Tell Stitch how to apply the results of your blocking strategies.
All records with scores that are equal to / better than this value cluster together.
Configure preprocessing profiles to support a variety of edge cases.
Amperity stores 30 days of changes to clusters and primary keys by default.
Configure Stitch reports to include or exclude specific Amperity IDs.
A threshold at which matching records do not identify unique individuals.
The threshold at which nearly-identical records collapse into one record.
