What is data lineage?
Lineage connects a number in a dashboard back through the joins, filters, mappings, and calculations that produced it, all the way to the source systems.
Think of it as a chain of custody for data. If a revenue figure looks wrong, lineage tells you which source fed it, which definition calculated it, and what changed recently, so you can fix the cause instead of arguing about the result.
How does data lineage work?
Lineage is captured as data moves through a pipeline. Each step records its inputs, the logic it applied, and its outputs. Linked together, those records form a map from source to report.
| Stage | What lineage records | Example |
|---|---|---|
| Source | Which system and account the data came from | Ad platform spend for one account |
| Ingestion | When it arrived and in what shape | Daily load, with the field names the platform used |
| Transformation | Joins, mappings, and cleaning rules applied | Platform campaign names mapped to a business category |
| Definition | How the metric was calculated | ROAS as revenue net of returns divided by spend |
| Consumption | Which reports, tools, and agents use it | An executive dashboard and an AI answer |
What is an example of data lineage?
Tracing a number that changed
A weekly report shows cost per acquisition jumped 18% overnight. Before anyone changes a campaign, an analyst follows the lineage of the metric.
The trace shows that one ad platform renamed a conversion field, so conversions from that platform stopped mapping into the total. The fix is a mapping update, not a media change, and it takes minutes instead of days.
Data lineage vs. data provenance vs. data catalog
These terms are related and often used together, but each answers a different question.
| Data lineage | Data provenance | Data catalog | |
|---|---|---|---|
| Question it answers | How did this data get here, and what changed it? | Where did this data originally come from? | What data do we have, and what does it mean? |
| Focus | The full path and every transformation | The origin and history of a record | An inventory with descriptions |
| Typical use | Debugging, impact analysis, audits | Compliance and trust in sources | Discovery and documentation |
Why does data lineage matter for reporting and AI?
Every report, plan, and AI answer inherits whatever happened to the data underneath. Without lineage, checking a number means rebuilding its history by hand, and small errors spread quietly into everything built on top.
- Faster debugging. Teams find the cause of a wrong number instead of guessing.
- Impact analysis. Before changing a definition, you can see every report and tool it affects.
- Audits you can pass. Finance, legal, and security teams can see how a number was produced.
- AI answers you can check. If an AI answer looks off, the team can follow it back to the source data and the definition it used.
How does Alli do this?
Alli keeps its data foundation deterministic and traceable. Alli Datasources brings data in through governed ingestion, and Alli Semantic Layer models how performance tables connect to campaigns, ads, accounts, and calendars, defines every KPI once, and maps platform naming through Categorizations. Definitions are version-controlled and changes are deployed through controlled workflows.
Because every dashboard, Data Explorer view, Ask Alli answer, and alert reads from those same definitions, every number in Alli traces back to a defined source.
Explore Alli Semantic LayerData lineage FAQ
What is data lineage in simple terms?
It's the history of a piece of data: where it came from, where it went, and what was done to it along the way, so you can trace any number back to its source.
What is the difference between data lineage and data provenance?
Provenance focuses on where data originally came from. Lineage covers the full path, including every system it passed through and every transformation applied to it.
Why is data lineage important for AI?
AI tools answer from the data and definitions underneath them. Lineage lets teams check an AI answer against its source, and find the cause when an answer looks wrong.
How do you maintain data lineage?
Keep transformation logic and definitions in governed, version-controlled systems rather than scattered spreadsheets, and record each step as data moves from source to report.
Last updated ‹ Back to the glossary