Case Study
Autotrader Gains a Clearer View of High-Intent Car Shoppers
Snowplow Identities
One User Across Devices and Logins, Resolved in Real Time
Identities links anonymous and logged-in activity across devices, sessions, and domains as events move through your Snowplow pipeline. Personalization, attribution, and AI agents work from one user instead of several anonymous visitors.
What Batch Identity Resolution Costs Your Business
Most identity stitching runs after data lands in the warehouse. Until it does, one user looks like several anonymous visitors, and every system acting in the moment sees only part of the journey. Personalization can't use what the user did before logging in, at the point it could still change what they do next. Snowplow Identities closes the gap at login, while the user is still on your site, so you can act on the whole journey in real time.
Anonymous
- Searches "London 5-star hotels" on mobile
- Browses luxury hotels at full price on desktop
- Compares amenities and checks availability
Real-time identity gap
Known
- Logs in and starts booking
- Abandons the booking
Resolved Before the Warehouse
Every event leaves the pipeline with a resolved Snowplow ID. Attribution, personalization, and AI agents see the current user without waiting for an overnight stitching job.
Built on Your First-Party Data
Identities links identifiers you already collect, from cookie and user IDs to hashed emails and custom fields. Every match is deterministic, with no third-party identity graphs.
Owned and Run in Your Pipeline
The identity graph is deployed with your Snowplow pipeline. You set the rules and Snowplow runs the graph, so there's no stitching code to maintain.
Autotrader replaced a daily batch job with millisecond identity resolution. With stitched profiles, its purchase-intent model correctly identified 82% of shoppers who went on to contact a dealer, up from 63%.
"The Snowplow Identities product has helped us move from thinking about devices to thinking about people."
Resolve Identity in the Stream, Not the Warehouse
Identities runs as part of enrichment. Each event is checked against the identity graph and leaves the pipeline with a resolved Snowplow ID. Your warehouse, Snowplow Signals, and event forwarding destinations all see the same user.
Attribute anonymous pre-login activity to the logged-in user, in the same session or across sessions and devices
Connect web, iOS, and Android activity when users log in with the same credentials
Send identity-resolved events to your marketing and engagement tools with Event Forwarding
Handle Edge Cases Without Custom SQL
Choose the identifiers that define a unique user for your business, mark the one that must never merge, and filter which events get resolved. The graph handles the shared devices, repeated logins, and cross-domain journeys that break hand-written stitching jobs.
Keep people who share a device separate by marking an identifier, such as user ID, as unique
Follow users across your domains with cross-domain tracking aliases
Resolve on standard identifiers or any field in your events, such as a hashed email or custom user ID
Keep Identity in Your Own Infrastructure
The identity graph is deployed in the same cloud as your Snowplow pipeline, with no third-party data sharing. On Private Managed Cloud, that's your own cloud account. Resolved IDs flow into your warehouse and into Snowplow Signals for real-time features.
Fulfill GDPR erasure requests through the Identities API, with a dry-run preview first
Build identity mapping tables in your warehouse with the Identities dbt package
Track graph health in Console, including deduplication rate and identifiers per profile
Identity Belongs in Your Real-Time Pipeline
When identity is resolved outside the pipeline, every system acting in the moment works from partial data until it catches up. Each alternative trades away speed, ownership, or engineering time.
| Approach | Where identity is resolved | The tradeoff | With Snowplow Identities |
|---|---|---|---|
| Packaged CDPs | In the vendor's platform, on data its own SDKs collect | Identity data lives in another vendor's system, which limits ownership, governance, and architectural flexibility | Identity is resolved inside your Snowplow pipeline, with no separate collection layer or vendor profile store |
| Warehouse-native CDPs | In the warehouse, after data loads | Identity updates on warehouse batch cycles, so in-session uses like personalization are difficult to support | Identity is resolved in the event stream, and the same resolved IDs land in your warehouse for analysis |
| Identity graph specialists | In the vendor's graph, often with third-party data | Built mainly for advertising, with matches that depend on external identity graphs | Deterministic matching on your own first-party behavioral data, for product, analytics, and AI use cases |
| In-house builds | In stitching jobs your team writes and runs | Your team owns the graph logic, edge cases, and maintenance indefinitely, competing with the product roadmap | Snowplow runs the graph and your team configures identifiers. Autotrader tested an in-house streaming build against Identities, then retired the build. |
