Enrichment Introduction

What is enrichment?

Enrichment is Qualytics writing its findings back into your own infrastructure. When a datastore is linked to an enrichment destination, every Scan writes its results and source record examples into that destination; Materialize and Export operations write there too. Without a destination, results exist only inside Qualytics. With one, they are queryable, auditable, and available to whatever else reads your data.

An enrichment datastore is where those findings land. It appears in the same tree as every other datastore, and the tree's Enrichment filter chip narrows the list to these datastores. A source datastore links to one as its enrichment destination, and everything written for that source goes there.

Keep these outputs in a dedicated location, separate from the source data you govern, even when both use the same connection. Create the enrichment datastore on a supported connector using Add > Enrichment on the Datastores page, Add new inside the link dialog, or the API with enrichment_only set to true. Then link the datastores that should write to it. The picker lists only datastores created for this purpose; it does not turn a source datastore into a destination.

Why link a destination?

• Keep the evidence. Source record examples are written to the destination, so you can inspect them with your own tools rather than only inside Qualytics.
• Build a history. With the Append remediation strategy, every Scan adds to a record of everything that ever failed, which compliance and governance work can rely on.
• Feed other systems. BI tools, pipelines, and alerting can read the written outputs and act on quality findings.
• Query across datastores. Several sources can write to one destination, so one query can cover enrichment data from all of them.
• Run Export and Materialize. Both operations write into the destination and are unavailable without one.
• Receive recipe outputs. The Entity Resolution Recipe writes its golden set and optional excluded-duplicates output to the destination as materialized outputs.

What happens without one?

• Scan findings remain inside Qualytics rather than being persisted in infrastructure you control.
• No source record examples are available for inspection outside Qualytics.
• No remediation outputs: the Append and Overwrite strategies are unavailable.
• No Export or Materialize operations: both return an error without a destination.

What gets written, and by what

Each container in a destination carries a purpose, shown as a badge, that says which operation wrote it: Scan, Remediation, Materialize, Export, Computed, or Other. What each holds, where the badge appears, and what a written container shows are on Enrichment Purposes.

By default, writes from Scan, Materialize, and datastore Export trigger Sync and Profile on the destination, so the new containers are ready to use. This can be turned off per source datastore; the separate Export Check Templates action requires manual Sync and Profile. The full mechanics, including settings, display names, lineage, and unlinking, are on How Enrichment Works.

What it can and cannot do

An enrichment datastore shares most capabilities with any other datastore:

• Computed assets. Create computed tables, files, and joins over the written containers; they carry the Computed purpose.
• Operations. Run Sync, Profile, and Scan on it like any datastore.
• Flows. Pick it for an operation action, and let its operations trigger flows, like any datastore.

Three things are deliberately not offered:

Enrichment datastores are excluded from estate-level quality insights. They do not inherit the Observability defaults, and their AI effort and draft-inference settings are left alone, including when Apply to existing is used. AI effort starts at Off when one is created, and the Profile settings remain editable on that datastore. Scan defaults, including the anomaly limits, still apply. See Platform Defaults.

Side effects of linking

• Every Scan writes. Scan results, source records, and scan metadata are written on every Scan, not just the first.
• Append accumulates. The Append strategy adds anomalous records after every Scan and never removes them; on large containers with recurring anomalies, storage grows quickly.
• Overwrite replaces remediation data. It keeps the latest remediation snapshot; the four Scan metadata outputs still accumulate history.
• Prefixes must be unique. Each source sharing a destination needs its own prefix; Qualytics refuses a link that would reuse one.
• Unlinking is Admin-only. Linking needs Member and Editor; unlinking needs Admin, because it stops all future writes.

Next Steps