New Features

Redshift Data Warehouse Support

Weld now supports Amazon Redshift as a destination data warehouse — both Serverless and Provisioned Cluster deployments. You can now land synced data directly into Redshift alongside BigQuery, Snowflake, and Postgres.

Explore Amazon Redshift documentation →

Primary Key Visibility on Incremental Syncs

Primary key columns are now displayed in the UI for most incremental sync streams. Primary key columns are locked to prevent accidental filtering or hashing transformations, protecting downstream data integrity.

Primary key visibility in incremental sync streams

Multi Account Backfill

Multi Account Backfill has been added to connectors that support multi-account selection, including Google Ads, Google Analytics, and Microsoft Ads. This enables syncing historical data across multiple selected accounts or properties at once. When you add a new account to an existing connection, Weld will automatically backfill historical data starting from the incremental pointer start date — so new accounts are always brought up to the same historical baseline as your existing ones.

Multi-account backfill selection

New Connectors

Reddit Ads

Full ELT support for Reddit advertising data including campaigns, ad groups, ads, and performance reporting.

Explore Reddit Ads documentation →

Reddit Ads connector setup

Reddit Ads ERD

To help you model Reddit Ads data faster, here is an overview of the core table relationships in the connector.

Reddit Ads ERD overview

The Reddit Ads ERD gives you a quick overview of how campaign, ad group, ad, and performance tables relate.

Explore full Reddit Ads data schema →

Connector Updates

Major reliability and scalability improvements:

  • Universal Soft Deletes: Soft deletes are now enabled for all time-based reports across all accounts. If records are modified or disappear within a synced date range, Weld will reconcile and soft-delete stale rows in your warehouse.
  • Multi-Account Syncing: Completely refactored to support per-account incremental pointers, massively improving scalability for agencies or teams syncing multiple Google Ads accounts.
  • Smart Notifications: Targeted warning notifications if Weld loses access to a specific Google Ads customer account, while the rest of your accessible accounts continue syncing normally.

Google Analytics 4

You can now select and sync data across multiple GA4 properties simultaneously. We also added multi-account backfill streaming and improved caching to drastically reduce Google API quota consumption.

Selecting multiple GA4 properties

Shopify & Shopify Plus

  • Abandoned Checkouts: Now uses chunked mode for faster, more reliable syncing.
  • Transactions: Added payment_id field.
  • Fixed logic for parsing staffMembers and updated OAuth token refresh handling to prevent disconnects.

Amazon Selling Partner

  • New Stream: Added fba_inventory_summary — a history-mode stream for FBA Inventory Summary.
  • Improved error handling so that Amazon's "Data Not Ready" errors on Search Query Performance reports no longer quarantine your entire sync.

AppLovin

  • New Stream: Added hourly_advertisement_report for hourly granularity in advertisement reporting.

Microsoft Ads

  • Fixed an issue with max retention dates on reports. Corrected the MS Ads number parser.

TikTok Ads

  • Optimized syncs to run lookbacks only once a day to save resources. Fixed reporting metrics.

Google Sheets

  • Upgraded error handling to provide clear, actionable messages in the UI when Google service accounts encounter "Invalid Grant" or permission issues.

Facebook & Instagram

  • Added a delay to Facebook post comment fetching to protect your app from hitting API rate limits.
  • Fixed a bug where missing expiresIn tokens from Instagram caused sync failures.

Chargebee

  • Added plan_id field to the subscriptions table.

AWS S3

  • Added more graceful error handling and detailed logs when an S3 source stream fails.

Core ELT & CDC Improvements

  • CDC Parser Upgrades: Automatic conversion of nanosecond timestamps to milliseconds. Intelligent boolean-to-integer mapping to match Avro schemas.
  • CDC UI Safeguards: Improved state management to prevent manual triggers while a large initial CDC snapshot is still running.
  • CSV Download Optimization: Refactored the CSV download engine to handle massive queries from BigQuery, Snowflake, and Postgres with drastically lower memory usage.
CDC overview in Weld

For the full technical details including schema changelogs, see the February 2026 Release Notes.