---
schema_version: "1"
id: 41418b94a93efc951b568c867a72f70a8d4eba977b53e69c31a3f5c1e2dc84d1
title: Defining a Property Graph
collection: developer
version: current
locale: en-us
portal_url: https://docs.salesforce.com/developer/current/en-us/data-cloud/data-cloud-query-guide/references/data-cloud-sql-reference/pgq/define-property-graph.md
canonical_url: https://developer.salesforce.com/docs/data/data-cloud-query-guide/references/dc-sql-reference/define-property-graph.html
source_site: developer.salesforce.com
content_type: documentation
content_hash: d7021d83f42e3be8bc9f39bfc347d46264cd50ab26807a8dd43efb4c7162de0d
products:
  - Data 360 Query Guide
  - Data 360 Developer Documentation
---
# Defining a Property Graph

**Applies to:** ✅ Data 360 SQL ✅ Tableau Hyper API

A property graph tells Data 360 SQL how your Data 360 objects connect.

- Customers, orders, and products become **vertices**.
- "This customer placed this order" and "this order includes this product" become **edges**.
- Selected object fields become **properties** that you can filter or return.

You must define the graph in the SQL statement that queries it. Graph definitions can't be persisted and referenced in other SQL statements.

```sql
WITH <graph_name> AS PROPERTY GRAPH (
    VERTEX TABLES ( <vertex_table>[, ...] )
    [ EDGE TABLES ( <edge_table>[, ...] ) ]
)
<select_using_graph_table>
```

`EDGE TABLES` is a synonym for `RELATIONSHIP TABLES`.

## What Each Clause Means

Use these clauses to turn Data 360 objects into graph elements.

### Vertex Table

```sql
<table> [AS <alias>] KEY ( <column>[, ...] )
    [ LABEL <label> [ <properties> ] ] ...
```

- `<table>` is the Data 360 object, such as `ssot__Individual__dlm`. It can also be a CTE or a parenthesized subquery when you need a filtered subset of records.
- `AS <alias>` gives the element a new name. The `REFERENCES` clause in an edge table definition uses the alias, not the original table name.
- `KEY` identifies each record. Use the object's ID field. A key can list more than one field when one field isn't unique. The listed fields must exist on the source and be distinct.
- `LABEL` is the name you use later in `MATCH`, such as `individual` or `product`. More than one table can use the same label. Repeat `LABEL` on the same table to give that table more than one label. Each `LABEL` on a table can list its own properties. If you omit `LABEL`, Data 360 SQL uses the table alias, or name if no alias is given.

### Edge Table

```sql
<table> [AS <alias>] KEY ( <column>[, ...] )
    SOURCE KEY ( <column>[, ...] ) REFERENCES <vertex_table> [ ( <key_column>[, ...] ) ]
    DESTINATION KEY ( <column>[, ...] ) REFERENCES <vertex_table> [ ( <key_column>[, ...] ) ]
    [ LABEL <label> [ <properties> ] ] ...
```

- `SOURCE KEY` is the field that points to the starting vertex. In the commerce graph, `ssot__SoldToCustomerId__c` points to the customer, and `ssot__SalesOrderId__c` points to the order.
- `DESTINATION KEY` is the field that points to the ending vertex. `ssot__Id__c` on the Sales Order points to that order vertex, and `ssot__ProductId__c` points to the product.
- The number of source or destination fields must match the referenced vertex key, and the types must be comparable. The optional column list after `REFERENCES` maps those fields to vertex key fields by position. The listed fields must be key columns of that vertex. If you omit the list, Data 360 SQL uses the vertex key in declared order.

If an edge's source or destination is `NULL` or doesn't match a vertex, Data 360 SQL leaves that edge out of the results. That isn't an error.

### Properties

```sql
PROPERTIES ( <expression> [AS <property>][, ...] )
PROPERTIES ARE ALL COLUMNS
PROPERTIES ALL COLUMNS EXCEPT ( <column>[, ...] )
NO PROPERTIES
```

Start with an explicit list of the fields you need, and give them short names such as `name` or `quantity`. Later queries use those property names, not the original `ssot__...` field names.

When you want every field, use `PROPERTIES ARE ALL COLUMNS`. Use `EXCEPT` to hide fields you don't want as properties, or `NO PROPERTIES` when the table is only a connection. If you expose a field without an alias, `MATCH` and `COLUMNS` use the original field name, such as `i.ssot__FirstName__c`.

## Example: An Ecommerce Graph

This graph answers "which products did this customer order?"

```sql
WITH pg AS PROPERTY GRAPH (
    VERTEX TABLES (
        ssot__Individual__dlm AS individuals KEY (ssot__Id__c)
            LABEL individual PROPERTIES (
                ssot__Id__c AS individualId,
                ssot__FirstName__c AS name
            ),
        ssot__SalesOrder__dlm AS orders KEY (ssot__Id__c)
            LABEL salesOrder PROPERTIES (
                ssot__Id__c AS orderId,
                ssot__Name__c AS name
            ),
        ssot__GoodsProduct__dlm AS products KEY (ssot__Id__c)
            LABEL product PROPERTIES (
                ssot__Id__c AS productId,
                ssot__Name__c AS name
            )
    )
    EDGE TABLES (
        ssot__SalesOrder__dlm AS customerOrders KEY (ssot__Id__c)
            SOURCE KEY (ssot__SoldToCustomerId__c) REFERENCES individuals
            DESTINATION KEY (ssot__Id__c) REFERENCES orders
            LABEL customerOrder PROPERTIES (
                ssot__CreatedDate__c AS createdDate
            ),
        ssot__SalesOrderProduct__dlm AS orderProducts KEY (ssot__Id__c)
            SOURCE KEY (ssot__SalesOrderId__c) REFERENCES orders
            DESTINATION KEY (ssot__ProductId__c) REFERENCES products
            LABEL orderProduct PROPERTIES (
                ssot__CreatedDate__c AS createdDate,
                ssot__OrderedQuantity__c AS quantity
            )
    )
)
SELECT * FROM GRAPH_TABLE(pg
    MATCH (i:individual)-[:customerOrder]->(:salesOrder)
        -[line:orderProduct]->(product:product)
    COLUMNS (i.name AS individualName, product.name AS productName,
        line.createdDate, line.quantity)
) AS g;
```

Read the `MATCH` clause as a walk: start at a customer, follow the order they placed, then follow the line item to the product.

### Expose Every Field (or Hide Some)

Use `PROPERTIES ARE ALL COLUMNS` when you want every field. Use `EXCEPT` to hide IDs. Use `NO PROPERTIES` when the table is only a connection. Fields exposed without aliases keep their original names, such as `i.ssot__FirstName__c`.

```sql
WITH pg AS PROPERTY GRAPH (
    VERTEX TABLES (
        ssot__Individual__dlm AS individuals KEY (ssot__Id__c)
            LABEL individual PROPERTIES ARE ALL COLUMNS,
        ssot__SalesOrder__dlm AS orders KEY (ssot__Id__c)
            LABEL salesOrder PROPERTIES ALL COLUMNS EXCEPT (ssot__Id__c),
        ssot__GoodsProduct__dlm AS products KEY (ssot__Id__c)
            LABEL product PROPERTIES ALL COLUMNS EXCEPT (ssot__Id__c)
    )
    EDGE TABLES (
        ssot__SalesOrder__dlm AS customerOrders KEY (ssot__Id__c)
            SOURCE KEY (ssot__SoldToCustomerId__c) REFERENCES individuals
            DESTINATION KEY (ssot__Id__c) REFERENCES orders
            LABEL customerOrder NO PROPERTIES,
        ssot__SalesOrderProduct__dlm AS orderProducts KEY (ssot__Id__c)
            SOURCE KEY (ssot__SalesOrderId__c) REFERENCES orders
            DESTINATION KEY (ssot__ProductId__c) REFERENCES products
            LABEL orderProduct PROPERTIES (ssot__OrderedQuantity__c AS quantity)
    )
)
SELECT * FROM GRAPH_TABLE(pg
    MATCH (i:individual)-[:customerOrder]->(:salesOrder)
        -[line:orderProduct]->(product:product)
    COLUMNS (i.ssot__FirstName__c AS individualName,
        product.ssot__Name__c AS productName, line.quantity)
) AS g;
```

## Related Documentation

- [Property Graph Queries (PGQ)](https://developer.salesforce.com/docs/data/data-cloud-query-guide/references/dc-sql-reference/pgq-context.md)
- [Querying with GRAPH_TABLE and MATCH](https://developer.salesforce.com/docs/data/data-cloud-query-guide/references/dc-sql-reference/graph-table-match.md)
- [Cross-Label Property Access](https://developer.salesforce.com/docs/data/data-cloud-query-guide/references/dc-sql-reference/cross-label-property-access.md)
- [WITH Clause](https://developer.salesforce.com/docs/data/data-cloud-query-guide/references/sql-commands/select-with.md)