Snowflake lets companies share a table with partners without copying it. ID providers, retail media partners, and ad platforms can read new rows as soon as they arrive, so there’s no replication lag and no ETL to maintain. That makes the warehouse a place where companies and their partners can work with the same data at the same time.

MetaRouter now supports Snowpipe Streaming as a Snowflake destination, available as a Starter Kit. Website, mobile web, and app events go from a brand’s first-party pipeline straight into a Snowflake table using Snowflake’s streaming API. There’s no virtual warehouse or stage to manage, and rows usually show up within seconds. It works alongside our existing batch Snowflake integration and doesn’t replace it.

‍

Shared on arrival means ready on arrival

If partners can read a row as soon as it lands, then whatever is in that row is what they see. There’s no staging step or nightly job where someone could catch an event that shouldn’t have been collected, or a record that isn’t linked to the right customer.

So the data needs to be right before it’s written. MetaRouter resolves identity when the event is collected, so each event already has the anonymous and known IDs that link it to one customer. Consent is applied server-side, inside the company’s own cloud, so an event without consent never gets into the table and never gets shared.

‍

The table is the contract

Snowpipe Streaming only accepts data that fits the table schema. If a field has no matching column, it’s dropped. If a value can’t be converted to the column’s type, the row is skipped. Neither causes an error, so the load still looks successful even though some data is missing.

MetaRouter formats events to fit the schema before they’re sent. Core fields go into typed columns, and properties, context, and traits go into VARIANT columns, so new properties aren’t lost and don’t need a schema change. Playbook transforms handle anything else that needs reformatting. If Snowflake’s ERROR_LOGGING setting is turned on, any skipped rows can be recovered.

‍

How it works

1. Collect: A first-party tag and mobile SDKs collect events from the brand’s own website and apps, with no vendor JavaScript on the page.
2. Resolve: Identity is matched at collection, from anonymous visitor to known customer, so every row is tied to a customer.
3. Enforce: Consent is applied server-side before anything is sent to Snowflake.
4. Map: Events are matched to columns by name, with typed columns for core fields and VARIANT columns for everything else.
5. Deliver: Events are streamed with offset-based delivery, so retries don’t create duplicates. MetaRouter connects using an RSA key pair the customer creates, limited to one table and pipe. Delivery can be checked in the MetaRouter monitoring dashboard and in Snowflake’s channel history.

‍

What changes for the business

When events arrive within seconds and already include identity and consent, partners can use a shared table the same day instead of the next morning. A retail media partner can build audiences from today’s sessions, and a clean room analysis can use current data instead of yesterday’s load. Because each event is defined once at collection, every team and partner counts a purchase the same way.

This runs on the same pipeline as every other MetaRouter destination, so the events in Snowflake are the same ones sent to Google, Meta, and The Trade Desk.

‍

Get started

Snowpipe Streaming for Snowflake is available now. Read the Snowflake Streaming documentation to set it up.

Not working with us yet? Talk to us. We'll show you what's possible inside your own cloud.

MetaRouter is the first-mile Customer Context Infrastructure that captures identity, behavior, and consent — in real time, inside the enterprise's own cloud, before the moment is lost. Trusted by global B2C enterprises including The Home Depot, Costco, Dicks Sporting Goods, and the world's largest Retail Media Networks. For more information visit: www.metarouter.io.

‍