Skip to main content
RudderStack is a customer data platform that collects first-party behavioral events and streams them to a warehouse. SourceMedium reads those events from RudderStack’s BigQuery warehouse destination and uses them in funnel and multi-touch attribution reporting. There is no API key. RudderStack writes event tables (and <event>_view views) into BigQuery; SourceMedium reads them. Plan requirement: New connections require Pro or Enterprise. Existing connections remain available on Foundation. See Foundation and Pro.
Sign in to SourceMedium, open your store, then choose Data sources → Connect data source and select the platform. You need an Admin or Editor workspace role.After you connect explains the first sync and connection statuses.

Vendor documentation

RudderStack’s destination asks for a Project, Location, Staging GCS Storage Bucket Name, Namespace (the dataset), and a service account Credentials file. You cannot change the namespace later. The Pro steps below are for whoever administers your Google Cloud project.

Plan requirements

To add a new connection on Foundation, ask a workspace admin to upgrade to Pro. In Connect, the source shows Requires Pro and links to your plan details. If your workspace already has this source connected, you can continue using and managing it on Foundation.

Pro: self-serve BigQuery destination

Pro delivers a dedicated BigQuery project you administer. Connect shows the project ID. Create the destination in RudderStack pointed at that project, then select and verify its tables in SourceMedium.

What you need

  • An active RudderStack workspace with a web or mobile source
  • Access to your dedicated BigQuery project (Connect shows the project ID)

How to connect

1

Create a dataset and staging bucket

In your dedicated project, create a dataset (for example rudderstack) and a Cloud Storage bucket in the same region. RudderStack stages files in the bucket before loading BigQuery.
2

Create a write service account

Create a service account with BigQuery Job User on the project, BigQuery Data Owner on that dataset (BigQuery Data Editor is enough if the dataset already exists), and Storage Object Creator plus Storage Object Viewer on the bucket. Download the JSON key. RudderStack’s BigQuery destination guide lists these roles and the equivalent custom-role permissions.
3

Add the BigQuery destination in RudderStack

In RudderStack, add a BigQuery destination on the sources you want to sync. Enter the project ID from Connect, the location, the bucket name, namespace rudderstack (or the dataset you created), and paste the JSON key into Credentials.
4

Confirm in SourceMedium

Select data for this store only. If the export contains multiple stores, first create a BigQuery view that filters to this store (for example, by your Snowplow app_id or RudderStack store property). SourceMedium assigns the selected rows to the store you are configuring, and each table can belong to only one store.After the first load, choose Find my datasets in SourceMedium. Select your dataset and event tables or deduplicated views. Map the event ID, event timestamp and loaded timestamp; both timestamps must be BigQuery TIMESTAMP columns. Optional columns supply visitor, session, page, order, revenue and campaign information. For sessions, map anonymous_id and context_session_id; SourceMedium combines them using the established visitor-session key. A context session ID without its anonymous visitor ID is not a complete session identity.Review the event mappings. With no event-name column selected, SourceMedium uses each table’s name. If your data uses one combined event table, choose its event-name column and enter the exact values, including capitalization. Map revenue only with the completed-order event. Conditional events, such as the shipping step of checkout_step_completed, need a view that gives that step its own event name.Choose Verify and save source. A successful check confirms read access and saves the mapping. It does not mean your events are in reports yet, and the connection shows Awaiting first data until SourceMedium confirms data from it. Existing historical events remain available. You can’t change the event ID or event timestamp column after saving, because that would duplicate events already in reporting. Contact SourceMedium to change them.

What data syncs

SourceMedium reads first-party behavioral events from RudderStack and maps the standard ecommerce events to its funnel events. For example: Where it lands in the warehouse: obt_funnel_event_history, the same table used for GA4, Snowplow, and other tracking sources.
  • Select RudderStack’s deduplicated <event>_view views when available. The reporting adapter also collapses repeated copies of the same event ID.
  • Only events that map to a SourceMedium funnel event are ingested. Micro-steps with no equivalent are not carried through.
  • RudderStack feeds funnel and attribution reporting, not order or catalog tables. Order data comes from your commerce platform integration.

Troubleshooting

  • Events not appearing? Confirm RudderStack’s BigQuery destination is active and the event tables and their <event>_view views are landing in the dataset you configured.
  • Missing a funnel step? Open Configure warehouse source from the connection and check the table, column and event mappings. Optional columns may be absent, but every selected table needs the required ID and timestamp columns.
  • Table too large to check? A view that would scan more than 10 GB can’t be checked. Select the underlying event table, or a view that reads less data.
  • Resumed a paused source? SourceMedium stops reading a source when it is paused. After you resume it, the connection shows Setting up until you open Configure warehouse source and choose Verify and save source. Events already in reporting stay.
  • Attribution looks incomplete? RudderStack captures first-party on-site events only. Combine it with your ad platform and commerce integrations for full multi-touch attribution.