Case study 04 / 06
Purely White Deluxe
Oral care · Marketplaces
Target Plus, wired straight into Shopify and Flexport
- Data engineering
- Software engineering
- Cloud & infrastructure

Key figures
- Target polled for new orders
- 5 min
- Reconcile as a safety net
- 15 min
The brief
The challenge
Selling on Target Plus means meeting Target's rules on every listing and every order (acknowledgements, exact shipping methods, cancellation reasons, stock per distribution center) while the catalog lives in Shopify and stock and shipping live at Flexport.
What we built
A direct Target Plus integration on the client's own Google Cloud: listings published from Shopify against Target's taxonomy; orders acknowledged and routed to Flexport; tracking, cancellations and inventory reported back to Target, and returns registered at Flexport, all operated from a dedicated console. Alongside it, a DataSync warehouse brings marketplace and order data into BigQuery.
Outcome
A marketplace integration built in the brand's own Google Cloud: automated tests, vendor contracts validated against Target's and Flexport's published API specifications, and one pipeline that builds separate sandbox and production apps from the same code.
Related services: Platform integrationsCloud infrastructure
Expertise
What it took.
Data engineering
- Incremental syncs of the Shopify catalog and Target listings and returns, with cursors and overlap windows.
- A transactional outbox with priorities, backoff and stall detection, so order, stock, price, listing and return calls survive a failure, and idempotency keys keep a retry from applying twice.
- Store-then-process webhooks with dedupe, a 15-minute reconcile as a safety net, and change-only inventory snapshots with a full ledger.
- DataSync: marketplace and order data landed in the brand's own BigQuery, including HMAC-verified order webhooks stored as append-only events.
Software engineering
- A FastAPI backend with an order state machine, idempotency keys and a circuit breaker on automatic cancellations.
- A listing publisher built on Target's data model: variation families, taxonomy-driven attributes and pre-send validation.
- Automated tests, including vendor contract mocks validated against Target's and Flexport's published OpenAPI specifications.
- An operations console with in-app help and hand-drawn diagrams, from the system map to the order, listing and return lifecycles.
Cloud & infrastructure
- One Cloud Build pipeline that provisions and deploys three Cloud Run services per app: UI behind Identity-Aware Proxy, IAM-only jobs, public webhook receiver.
- Cloud SQL Postgres behind a PgBouncer VM on a private VPC, Cloud Scheduler with OIDC, and Secret Manager key rotation without redeploys.
- Cloud Monitoring alert policies, in-app alerts and an insert-only database role for the public webhook.
Project 01
Target Plus middleware
FastAPI · React 19 · Cloud SQL · Cloud Run
Publishes the Shopify catalog to Target Plus (variation families, Target's taxonomy, list, offer and MAP prices, stock per distribution center), then polls Target for new orders every five minutes, acknowledges them, routes them to Flexport, reports tracking, cancellations and inventory back to Target, and registers Target returns at Flexport for inspection and restock. Everything is operated from a console with an attention queue and a full audit trail.
- Goal
- Meet Target's operating rules automatically on every listing and order, without ever losing or duplicating an order when a vendor API times out.
- Highlights
- Transactional outbox with priorities, backoff and stall detection
- Store-then-process webhooks plus a 15-minute reconcile
- Circuit breaker on automatic cancellations
- Vendor contract mocks validated against Target's and Flexport's API specifications
- Stack
- Python
- FastAPI
- React 19
- PostgreSQL
- Cloud Run
- Cloud SQL
- Cloud Build







Project 02
Marketplace data warehouse
DataSync · BigQuery · Cloud Run
The brand's marketplace and order data lands in its own Google Cloud project through DataSync: a scheduled, signed TikTok Shop connector onboarded through a one-time seller link, and an HMAC-verified order-webhook receiver that appends every event to BigQuery with a latest-state view.
- Goal
- Bring the brand's marketplace and order data into a warehouse it owns and pays for.
- Highlights
- Per-resource sync configuration: staging table, merge target and keys
- Webhook events stored append-only, so schema changes never break ingestion
- Stack
- DataSync
- BigQuery
- Cloud Run
- Cloud Scheduler
- Webhooks


