How to Use Event Triggers in BigQuery

What are event triggers in BigQuery?

BigQuery itself does not run procedural triggers, but every job (INSERT, UPDATE, LOAD) emits Cloud Audit Logs. By creating a log sink and wiring it to Pub/Sub or Eventarc, you get an "event trigger" that invokes Cloud Functions or Cloud Run whenever a table changes.

Why use event triggers for ecommerce analytics?

Triggers update dashboards instantly, restock products when Orders spike, or email customers after an Orders insert—all without polling.

How do I create an event trigger for table inserts?

1️⃣ Create a Pub/Sub topic.
2️⃣ Add a log sink that filters BigQuery tabledata.insertAll events and routes them to the topic.
3️⃣ Deploy a Cloud Function subscribed to the topic.

Cloud Logging filter

resource.type="bigquery_resource" AND
protoPayload.methodName="tabledata.insertAll" AND
resource.labels.table_id="Orders"

Node.js Cloud Function skeleton

exports.ordersInserted = (msg, ctx) => {
const data = Buffer.from(msg.data, 'base64').toString();
const event = JSON.parse(data);
const rowCount = event.protoPayload.serviceData.tableDataChange.rowInsertCount;
console.log(`Inserted ${rowCount} rows into Orders`);
};

How do I test the trigger?

Run the example INSERT below. Watch Cloud Functions logs to confirm the message appears.

Best practices for production

• Use the narrowest possible log filter to cut costs.
• Grant least-privilege IAM roles (roles/viewer for the sink writer; roles/pubsub.subscriber for the function).
• Keep function runtime idempotent to handle duplicate events.

Common mistakes

Missing IAM roles – The sink service account needs Pub/Sub Publisher; the Cloud Function needs BigQuery Data Viewer if it queries tables.

Overly broad filters – Filtering only on resource.type="bigquery_resource" floods Pub/Sub and raises costs. Always include methodName and table/dataset labels.