Skip to main content
The Datagraph is the part of the DataHub where you describe how your Datasources relate to each other. It lets your Workflows reuse the same trusted data.
The Datagraph is in beta. Its pages and options can still change. Talk to your Technical Account Manager before you rely on it in a live Workflow.

​
Why Use the Datagraph?

Each Datasource holds one set of data — products, stores, purchases, or customers. On its own, a Datasource cannot answer a question that needs two sets of data at once. The Datagraph links Datasources together. Once linked, your data team can prepare reusable Entities (saved questions such as “top five products bought by this person”) that any Workflow can reuse.
You do not have to build the Datagraph yourself. It is usually set up once by your data team or your Reelevant Technical Account Manager, then reused by everyone.

​
Where to Find It

Open DataHub in the sidebar, then choose Datagraph. You will see three pages: On the Schema page, the Tables panel on the left lists the Datasources in the model. Selecting a table shows its Columns and its Relations (the links to other tables). Datagraph Schema page for a property listings model, with five SeLoger tables on the left and the columns of the selected users table on the right The Relations tab shows which other tables the selected table is linked to, and on which fields. Relations tab of a Datagraph table, showing two links from the recommendations table to the users table and the listings table

​
Understanding the Explorer

The Explorer is the most useful page if you simply want to understand your data.
  1. Open DataHub → Datagraph → Explorer.
  2. Stay on the Diagram tab to see every Datasource as a box, with lines showing how they are linked.
  3. Drag the diagram to move around, and use the zoom buttons to get closer.
  4. Open the Relations tab to review each link one by one.
  5. Open the Datasources tab to list the Datasources used in the model, with their fields.
Datagraph Explorer Diagram tab, showing five linked property Datasources as boxes connected by lines
If the Explorer asks you to create a Schema first, the Datagraph has not been set up yet. Ask your data team or your Technical Account Manager to build it.
On the Relations tab, each link shows a Join match percentage. It tells you how well the values on both sides of the link actually overlap in your data. Relations tab of the Datagraph Explorer, listing five links with their join match percentages
A weak or invalid link usually means the two Datasources do not share the same identifiers. Personalisation built on that link may return no result, so raise it with your data team before you use it in a Workflow.

​
Draft and Published Versions

The Schema and each Entity have their own version, shown as a status. Nothing you see in a Draft affects your live personalisation until someone publishes it. Datagraph Entities page listing five Entities with their Draft or Live status

​
What Entities Can Answer

An Entity answers one question about one person, one product, or one area. Your data team writes it once, and every Workflow can reuse it. Here are typical examples for a property listings account: The same patterns work for other businesses. A retail account might ask for the last items a person bought, the stock left in their favourite store, or the best sellers of their preferred category.
Describe the question you need in plain words, with the data it should return. Your data team turns it into an Entity, then you pick it in your Workflow.
On an Entity page, the Preview & results panel shows a sample of the answer, so you can check the data before it goes live. Datagraph Entity page showing the entity name, the editor, and two matching listings in the Preview and results panel The Input / Output panel lists the Parameters the Entity expects, and the fields it returns. Input and Output panel of a Datagraph Entity, showing the user_id parameter and the returned fields with their types

​
Using an Entity in a Workflow

Once an Entity is published, it can be used in a Workflow with the Query an entity Data Node (a Node that fetches external data for the Workflow).
  1. Open your Workflow in the Workflow editor.
  2. Add the Query an entity Data Node.
  3. Select the published Entity you need.
  4. Fill in the Entity parameters, either with a fixed value or with a Workflow variable.
  5. Use the returned data in your Content or in a Condition.
See Query an Entity for the full Node walkthrough.
Only users with the tech admin role can configure this Node. Ask your administrator if the Node appears read-only. See Permissions for the Datagraph Schema and Datagraph Entity permissions.

​
What’s Next?

DataHub Overview

Learn how Datasources are imported and configured.

Datagraph Deep Dive

Understand the data model, versions, and Entities in detail.

Explorer

Read a join match analysis and fix weak links.

Permissions

See who can view, edit, and publish Datagraph objects.