Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.upstackdata.com/llms.txt

Use this file to discover all available pages before exploring further.

Upstack Data is a five-stage pipeline that captures raw storefront events, resolves visitor identities, enriches events with customer context, forwards high-quality signals to ad platforms, and stores everything in an analytics warehouse you can query.
Understanding the pipeline architecture helps you diagnose where data quality issues originate and how each Upstack product contributes to your marketing data stack.

Why It Matters

Most marketing data stacks are assembled from disconnected tools — a pixel here, a CAPI integration there, a separate analytics platform. Each tool operates on its own copy of the data, with its own identity model, its own event schema, and its own gaps. Upstack replaces that patchwork with a single pipeline. Every event flows through the same sequence of stages, accumulating identity and context at each step. This means your conversion APIs receive the same enriched events that your attribution models analyze — no reconciliation, no duplicate schemas, no conflicting identity logic. A unified pipeline also makes debugging straightforward. If a conversion isn’t appearing in Meta, you can trace the event from capture through resolution, enrichment, and forwarding to pinpoint exactly where the signal was lost.

How It Works

The pipeline has five stages, each handled by a dedicated Upstack product.

1. Capture — Upstack Pixel

Shopper interactions on your Shopify storefront generate events (page views, add-to-cart, checkout, purchase). The Upstack theme extension sends these as first-party requests to Cloudflare Workers at the edge. Bot traffic is filtered, a durable first-party cookie is set, and events are normalized into a canonical schema. Latency is sub-50 ms with zero impact on storefront performance.

2. Resolve — Upstack ID

Raw events enter the AWS backend where the identity resolution engine matches anonymous sessions to known customers. Deterministic matching (email, phone, login) creates hard links. Session stitching retroactively attributes anonymous events once a visitor identifies themselves. Cross-device linking merges profiles when the same identifier appears on multiple devices.

3. Enrich — Upstack Enrichment

Resolved events are augmented with customer metadata from Shopify (order history, LTV, customer tags), geographic context from IP geolocation, and all available identity fields from the identity graph. After this stage, events carry the full set of match keys needed for high-quality forwarding.

4. Activate — Upstack Signal & Upstack Flows

Enriched events are queued via Amazon SQS and delivered server-to-server to configured destinations: Meta CAPI, TikTok Events API, Google, Klaviyo, Pinterest, Snapchat, and others. Upstack Flows handles Klaviyo-specific flow recovery. Each destination adapter transforms events into the platform’s required schema. Delivery includes automatic retry with exponential backoff and dead-letter queues for failed events.

5. Understand — Upstack Analytics

Every event is also written to SingleStore, a high-performance analytics database. Upstack Analytics queries this warehouse to build attribution reports, calculate ROAS by channel, and surface customer journey insights — all from the same identity-resolved, enriched events that your destinations received.

Infrastructure Summary

LayerTechnologyRole
EdgeCloudflare WorkersEvent capture, bot filtering, cookie management
ComputeAWS LambdaIdentity resolution, enrichment, forwarding, API handlers
QueueAmazon SQSAsync event delivery with retry and dead-letter handling
Identity storePostgreSQL, NeptuneIdentity graph storage and traversal
Analytics warehouseSingleStoreHigh-speed analytical queries for attribution and reporting
ConfigurationDynamoDBDestination configs, account settings, subscription data

Key Terms

TermDefinition
Edge layerThe Cloudflare-based infrastructure that captures events close to the visitor for low-latency, high-reliability collection.
Canonical schemaThe standardized event format used internally across all pipeline stages.
Dead-letter queueA holding queue for events that fail delivery after all retry attempts, preserving data for debugging.
Identity graphThe data structure mapping all known identifiers to unified customer profiles.
Match keysHashed identity fields attached to events that ad platforms use to link conversions to user profiles.

Server-Side Tracking

Deep dive into the Capture stage.

Identity Resolution

Deep dive into the Resolve stage.

Data Enrichment

Deep dive into the Enrich stage.

Conversion API Forwarding

Deep dive into the Activate stage.

Attribution & Reporting

Deep dive into the Understand stage.

Install Upstack Pixel

Get started with the first pipeline stage.