As an account grows, it gets harder to answer two questions with confidence: "where does this number actually come from?" and "what else will break if I change this dataset or column?" Without a way to see the path data takes, both require digging through datasets, merged datasets, and metric definitions by hand.
The lineage explorer maps that path for you. Starting from any dataset or metric, it draws a graph of everything upstream (what it's built from) and downstream (what's built on top of it) — data sources, datasets, merged datasets, and metrics — so you can trace a value to its source or check the downstream impact of a change before you make it.
From a dataset
- Go to Data Manager > Datasets.
- Find the dataset you want to trace and click the drop-down arrow (
) to open the options menu, or open the dataset and use the header actions. - Select Lineage explorer.
From a metric
- My Metrics — Click the three dots (
) on the metric card or row and select Lineage explorer. - Custom Metrics — Click the drop-down arrow () and select Lineage explorer.
- Metric Insights — Open the metric, click the three dots (), and select Lineage explorer.
The lineage explorer opens with your starting dataset or metric selected, its details open on the right, and one hop visible in each direction: upstream on the left, downstream on the right. Arrows always point in the direction data flows.
Each type connects to a different set of neighbors:
| Selected asset | Upstream (left) | Downstream (right) |
|---|---|---|
| Data source | — | Its datasets, custom metrics, and basic metrics |
| Dataset | Its data source | Merged datasets that use it, and custom metrics built on it |
| Merged dataset | Its input datasets | Further merged datasets, and custom metrics built on it |
| Basic metric | Its data source | Custom metrics that use it |
| Custom metric — from a data source or dataset | The dataset or data source it's built on | Custom metrics that use it |
| Custom metric — with calculations | Every metric in its formula | Custom metrics that use it |
The graph is read live, so it always reflects the current state of your account.
- Click a card to select it and open its details. The card and its direct neighbors stay highlighted; everything else fades.
- Click a +N badge to load the neighbors on that side that aren't shown yet.
- Click Explore from (top left) to pick any dataset, merged dataset, or metric already on the graph as the new origin. Select as origin in a card's options menu does the same.
- Use the zoom control (bottom right) to zoom in or out, or scroll to pan the canvas. Pinch or Ctrl + scroll also zooms.
- Press Esc to clear the current selection or column trace; press it again to close the lineage explorer. Clicking an empty part of the canvas also clears the selection.
Cards can't be dragged or rearranged, and there's no minimap, legend, or search box — the graph is meant to be explored one hop at a time from a starting point, not searched.
Every card in the graph has the same structure:
- Header — the asset's name, its integration logo or type icon, and a colored type label. Its options menu sits on the right.
- Status icon — a red warning means the connection has an error or is on hold; hover over it to see the error title and description. Otherwise, a verified badge can appear; hover to see who verified it.
- Rows — datasets list their columns, each with a type badge; calculated columns carry an
icon. Metrics list their definition, one row per part (see below). Five rows show at first; click Show more to see the rest. - +N badges — on the left or right edge of a card, showing how many neighbors on that side aren't loaded yet.
Expanding a metric card shows how it's defined, one row per part that's set. A metric with no definition rows can't be expanded.
| Row | What it shows |
|---|---|
| Measure | The aggregation and the column it's applied to, e.g. SUM(Revenue). If only one is set, only that one shows. |
| Datetime | The date column the metric is tracked by, or "Use sync date & time" if it uses the sync time instead. |
| Dimensions | Every column the metric can be broken down by. |
| Filtered by | The columns used in the metric's filters. Filter conditions and values aren't shown. |
| Formula | Custom metrics only — the formula, with each metric shown by name. A reference that can't be resolved, such as a deleted metric, stays as its raw token. |
How much of a metric's definition shows up depends on its type:
- Custom metric built from a dataset: shows its full definition.
- Custom metric built with calculations: shows its formula.
- Custom metric built directly on a data source: shows a partial definition, depending on the integration.
- Other integration metrics: show a header only.
Each card's options menu offers Select as origin, which makes that card the graph's new starting point, and View dataset, View data source, or View metric insights, which opens the asset's own page. Verify and Unverify are available on metrics and datasets.
Selecting a card opens its details on the right. Rows with no value are hidden, so the panel varies by asset.
- Metric — Measure, datetime, dimensions, filtered by, formula (custom metrics), verified by, verified at
- Column — Description, synonyms, unique values, fill rate, null count, top value
- Dataset — Description, synonyms, default time dimension, entity, rows, size, columns, last sync, verified by, verified at
- Merged dataset — All dataset rows, plus merge type, operator, and join conditions
- Data source — No detail rows of its own. Open View data source for status, sync interval, and owner
The panel's footer links to the asset's own page — View dataset, View data source, or View metric insights — and the lineage explorer is still open when you come back.
Click a column row on a dataset or merged dataset card to trace it through the graph. The trace lights up the column's full path and fades everything else, expanding any card along the way so its relevant rows are visible.
For a merged dataset, the trace follows the column mapping back to the input column it originated from, even if the column has a different name. It continues through any earlier merged datasets in the lineage.
For a regular dataset, the trace follows the connection back to the data source and stops there because data sources do not have column lists.
The trace follows the column into any merged datasets and custom metrics that use it.
The trace shows the merged datasets into which the column is mapped and continues through any subsequent merged datasets.
For custom metrics built directly from a data source or dataset, the trace highlights every metric that uses the column as a measure, date, dimension, or filter. The corresponding row is highlighted to show how the column is used.
The trace also includes custom metrics with calculations that reference those metrics, including metrics built from other calculation-based custom metrics. Because formulas reference metrics rather than columns, these appear as highlighted cards instead of highlighting a specific row.
To end the trace, click the selected column again, press Esc, or click an empty area of the canvas. To trace another column, select it instead.
- A dataset is visible if its data source is shared with you, or you created the dataset.
- A merged dataset is visible only if all of its input data sources are shared with you.
- A basic metric is visible if its data source is shared with you.
- A custom metric built from a data source or dataset is visible if that source is shared with you.
- A custom metric built with calculations is visible only if every metric in its formula is visible to you.
- The graph loads one hop at a time. The full chain never loads at once, so expand it with +N badges as needed.
- A deleted or hidden metric that a custom metric still depends on appears as a bare placeholder card, and its name may show as a raw token in the formula.
FAQ
Does the lineage explorer show where a metric is used?
Not currently. The graph traces a metric back to its source data and forward to any custom metrics built on it, but it doesn't show which analytics assets reference that metric.
Why does a +N count stay the same after I click it?
The count on a +N badge includes assets you don't have access to. Once you click it, only the assets you're permitted to see load in, so the badge's original number doesn't necessarily match what appears.
Why is a dataset or metric missing from the graph?
This is usually a permissions issue. A merged dataset only appears if you have access to all of its input data sources, and a custom metric only appears if you have access to every metric in its formula.
