Most data quality discussions focus on the state of data at a point in time: run an audit, find the problems, fix them. But in real systems, data is constantly changing — new records are added, existing records are updated, some are deleted. Tracking those changes as they happen is what Change Data Capture is designed to do.
What Is Change Data Capture?
Change Data Capture (CDC) is a set of techniques for identifying and tracking changes made to a database — specifically, insertions, updates, and deletions — and making those change events available for downstream processing.
Instead of periodically exporting the full state of a database and comparing it to the previous export, CDC captures each change at the moment it happens. The result is a stream of change events: "Record 1047 was updated at 14:32:07 — the email field changed from john@old.com to john@new.com."
CDC is used primarily in:
- Real-time data pipelines — Sync changes from an operational database to a data warehouse without full re-loads
- Event-driven architectures — Trigger actions (send an email, update a dashboard, sync to another system) when data changes
- Audit logging — Maintain a complete history of every change made to a record, including who made it and when
Sohovi validates your dataset before it enters the warehouse — catching format errors, nulls, and duplicates at the source.
How CDC Works (The Main Approaches)
Log-based CDC — The most reliable and lowest-impact method. Reads the database's transaction log (every database writes one) to capture change events without querying the database directly. Used by tools like Debezium, Fivetran, and Airbyte.
Trigger-based CDC — Database triggers fire when a record is inserted, updated, or deleted, and write the change event to a separate audit table. More portable than log-based, but adds write overhead to every operation.
Timestamp-based CDC — Periodic queries check for records with an updated_at timestamp newer than the last check. Simple to implement, but misses deletions and can have gaps if clocks drift or updates happen faster than the check interval.
Full table comparison — Export the full table, compare to the previous export, identify differences. Works for small datasets, extremely inefficient at scale.
Why CDC Matters for Data Quality
CDC is relevant to data quality in three specific ways.
1. Freshness and timeliness. One of the core data quality dimensions is timeliness — whether data is current enough for its intended use. CDC enables near-real-time data freshness in downstream systems by propagating changes as they happen, rather than waiting for nightly batch loads.
Sohovi measures all 10 data quality dimensions — completeness, validity, uniqueness, accuracy, consistency, and more — automatically across every column.
2. Audit trails and accountability. A CDC log is a complete history of every change to a dataset. If a record's address was changed incorrectly at 2pm last Tuesday, you can see that in the CDC log. This makes data quality investigation dramatically faster — instead of asking "how did this get wrong?" you can ask "show me the last three changes to this record."
3. Detecting anomalous changes. A sudden spike in update events, a pattern of deletions in a table that rarely has deletions, or a burst of changes to a field that's usually static — these patterns can indicate data quality problems at the source (a bad import, a migration gone wrong, user error at scale).
CDC and Data Quality in Practice
For most small businesses, full CDC infrastructure is not necessary. The concept most relevant to you is the audit trail: whenever critical records change, log who changed them, what changed, and when.
In spreadsheet-based workflows, this is as simple as maintaining a change log tab. In database-backed systems, most platforms have built-in audit features. In modern CRMs and SaaS tools, change history is usually available natively.
Sohovi finds gaps, duplicates, and format errors in your CRM data — so your team is working from records they can trust.
The principle is: don't just know the current state of your data. Know how it got there.
When Do You Actually Need CDC?
CDC infrastructure (log-based replication, streaming pipelines) is needed when:
- You have operational databases that feed analytics systems and you need near-real-time data
- You need a complete audit trail for compliance reasons (GDPR, HIPAA, SOX)
- You're building event-driven workflows that react to database changes
For most small businesses operating with CRM tools, spreadsheets, and SaaS platforms: your data change history is already captured in those tools. The discipline is to actually check it when data quality problems arise, rather than only examining the current state.
The Bottom Line
Change Data Capture is how serious data systems track the "what happened and when" of their data. Understanding CDC helps you appreciate why data quality degrades over time (changes accumulate without oversight) and why audit trails are essential for investigating and fixing quality problems when they appear.
