01
BigQuery pipeline
Native streaming sink to a BigQuery dataset of your choice. Handles schema migrations, late-arriving events and idempotent de-duplication automatically. Backfills if the warehouse goes down, no data lost.
Operator dashboards answer "what happened today". Warehouse pipelines answer everything else. SabNode streams every message, conversation, flow and contact event into BigQuery, Snowflake or Postgres, with CSV exports for one-off questions. Your data, modelled by your analysts, with no vendor lock on the spine.
Every serious analytics practice eventually hits the same wall: the vendor dashboards are fine for ops, but the moment leadership wants real cohort retention, LTV by channel, or a year-over-year revenue waterfall, you need the raw events in a warehouse where your analysts can model them. Most vendors make this difficult — opaque APIs, partial event coverage, schemas that change without warning, or premium pricing for the privilege of getting your own data out.
The failure modes compound. A growth analyst spends two weeks reverse-engineering a vendor's data export only to discover that delivery webhooks are missing entirely. A finance team builds a revenue report off CSV exports that silently drop the last 200 rows because the export ran during a deploy. A data engineer is asked to model attribution but the event timestamps are in three different timezones because nobody documented the schema. Six months later, the dashboards in the warehouse disagree with the dashboards in the vendor tool, and nobody trusts either.
The fix is to treat the warehouse export as a first-class product, not an afterthought. Document the event schema with version history. Stream events in near real time. Cover every event the platform emits — messages, conversations, flow steps, payments, opt-ins, CSAT — with consistent IDs across all of them. Then your analysts model once and the answers stay defensible quarter after quarter.
SabNode Exports & Warehouse is a first-class data product. Every event the platform emits — message sent, delivered, read, replied, failed; conversation opened, assigned, resolved; flow node entered, succeeded, dropped; contact created, updated, opted-in, opted-out; payment created, succeeded, refunded — is captured in a documented event schema with stable IDs that join across event types. The same contact ID appears in messages, flows, payments and CSAT, so cohort modelling works without manual joins.
Pipelines write to BigQuery, Snowflake or Postgres on a configurable cadence — hourly batch for most teams, near real time (1–5 minute latency) for high-velocity operations. Each warehouse target gets its own ingestion service that handles schema migration, late-arriving events and de-duplication idempotently. If your warehouse goes down for a few hours, no data is lost; the pipeline backfills automatically when connectivity returns.
CSV exports cover the rest. Any dashboard view, any list of conversations, any segment, any campaign report can be exported as a CSV with one click. Large exports (hundreds of thousands of rows) run as a background job and email a download link when ready. Exports are signed and short-lived to prevent accidental sharing of customer data, and every export is logged in the audit trail for compliance.
Schema is documented and versioned. The event schema is published as a versioned reference doc — every column, type, nullability and meaning, with example values and changelog entries when a field is added or deprecated. Breaking changes ship behind a major version bump with a migration window, so your warehouse models never break overnight. This is the difference between a data integration and a moving target.
Native streaming sink to a BigQuery dataset of your choice. Handles schema migrations, late-arriving events and idempotent de-duplication automatically. Backfills if the warehouse goes down, no data lost.
Snowflake pipe via Snowpipe or scheduled COPY commands. Lands events in a versioned schema with predictable column names and stable IDs that join across messages, conversations, flows and payments.
Postgres ingestion for teams running their warehouse on Postgres or self-hosted RDS. Supports change-data-capture style updates with deterministic upsert semantics, plus standard append-only event streams.
Choose the cadence per pipeline — hourly batch for cost-efficient bulk modelling, near real-time (1–5 min) for operational dashboards that drive same-day decisions, or both running in parallel against different schemas.
Any view, list or campaign report exports to CSV. Small exports download immediately; large exports (>50k rows) run as a background job and email a signed, short-lived download link when ready.
Every event, column, type and meaning is documented with example values, nullability rules and a changelog. Schema versions are stable; breaking changes ship behind a major version with a migration window.
Every export and every pipeline run is logged with user, timestamp, row count and purpose. Useful for DPDP, GDPR or SOC 2 audits where data movement must be traceable. Pipelines support data-residency rules per region.
A D2C analyst joins WhatsApp engagement events with Shopify order events in BigQuery to model lifetime value by acquisition channel. The shared contact ID makes the join trivial and the LTV-by-channel chart drives a 30 percent rebalance of marketing spend.
A B2B SaaS analyst models trial-to-paid cohort retention by onboarding-flow version. Snowflake export lets them slice retention by every flow variant they shipped that quarter and prove which onboarding flows actually retain users.
An edtech finance team builds a daily revenue waterfall in their warehouse — sign-ups, conversions, refunds, MRR change — joined with WhatsApp campaign exposure. The single source of truth ends a long-running dispute between marketing and finance.
An NBFC exports every customer message and opt-in record for a regulator audit. The platform produces a signed, scoped CSV bundle within minutes, with audit log entries that the regulator can verify independently.
A logistics operator streams driver-shift WhatsApp acknowledgements into Postgres alongside operational data. The combined dataset feeds a forecasting model that predicts roster gaps a week out, reducing missed deliveries.
Exports & Warehouse is included on every SabNode workspace. No separate billing, no extra setup — flip it on from your workspace settings.
Choose BigQuery, Snowflake, Postgres or CSV. Authenticate via OAuth or a service account scoped to the target dataset.
Pick which event categories to stream — messages, conversations, flows, payments, contacts, CSAT. Each becomes its own table in the destination.
Choose hourly batch or near real-time streaming. Real-time uses change-data-capture; batch uses scheduled bulk inserts. Both are idempotent.
Run a dry export. Validate that the schema matches your warehouse expectations and that example rows look right. Documented schema reference is shared.
Enable the pipeline. Monitor row counts, lag and errors in the pipeline health view. Backfills run automatically if the warehouse is unavailable.
Can't find what you're looking for? Talk to our team.
No credit card. No sales call required. Spin up a workspace, plug in a number, and your team is live in under an hour.