Digital Analytics
Data Model Pack
The Snowplow Digital Analytics Data Model Pack is a set of dbt packages that turn your Snowplow event data into analysis-ready tables in your own warehouse. Your analysts, BI tools, and AI agents query the tables directly.
Skip weeks of modeling raw events. Get web and mobile analytics, funnels, and attribution you can audit, on data you own.
- Web and mobile
- SQL you can read on GitHub
- Runs in your warehouse
What's included in the Digital Analytics Data Model Pack?
The pack is three Snowplow dbt packages. Each one runs incrementally in your warehouse and builds derived tables from your Snowplow events, so your team starts from modeled data instead of raw events. Snowplow maintains the packages, and the SQL is on GitHub, so your team can see exactly how every metric is calculated and extend it.
| Package | What it builds in your warehouse | Questions it answers |
|---|---|---|
| Unified Digital | Page and screen views with engaged time and scroll depth, sessions, users, conversions, and a user mapping table for stitching identifiers. Optional modules model consent, Core Web Vitals, and app errors. | Where does our traffic come from? How engaged are visitors? Which sessions convert? How many active users do we have? |
| Attribution | Paths to conversion, channel and campaign attribution under first-touch, last-touch, position-based, and linear models, a path summary, and an overview that shows return on ad spend once you add spend data. | Which channels and campaigns drive conversions? How does credit shift between models? What's our ROAS? |
| Media Player | Media plays, stats for each video or audio item, session and user-level media stats, and ad views, across web, mobile apps, and Roku. | Which content gets played, and how far do viewers get? How do viewers respond to ads? |
Extend it with the Ecommerce Analytics Data Model Pack to add cart, checkout, product, and transaction tables from the same Snowplow events.
What can you do with Snowplow's digital analytics data models?
The tables are a starting point for the analysis your teams already ask for. Here's what customers build on them most often.
Run funnel analysis to see where users drop off
Define the conversions that matter to you, from a sign-up to a visit to a specific page, and see how conversion rates change by channel, device, and campaign. You can change a conversion definition later and remodel past data with it.
Built on: Unified Digital sessions, views, and conversions
Run multi-touch attribution in your warehouse
See how credit for each conversion moves between channels and campaigns under first-touch, last-touch, position-based, and linear models. Add spend data from your ad platforms to calculate return on ad spend. The logic runs as SQL in your warehouse, so anyone can check how credit was assigned.
Built on: Attribution
Report on acquisition with your own channel rules
The default channel grouping draws on GA4 source categories, so reports look familiar from day one. Your team can override it with its own rules for UTM parameters and custom campaign fields.
Built on: Unified Digital views and sessions
Give every team the same numbers in your BI tool
The views, sessions, and users tables are ready for dashboards and self-serve analysis, so marketing, product, and analytics teams report from one definition of a session, a user, and a conversion.
Built on: Unified Digital
See which content holds attention
Rank pages and screens by engaged time and scroll depth, not just views, to see what visitors actually read and use.
Built on: Unified Digital views
Track active users and how often they come back
Define an active user your way and measure how often users return. A user mapping table links anonymous identifiers to logged-in user IDs, so sessions on different devices can roll up to one user.
Built on: Unified Digital users and user mapping
Measure video and audio engagement
Measure plays, play time, and progress for each video or audio item, plus ad views, across web, mobile apps, and Roku.
Built on: Media Player
Also included
- Consent reporting: consent rates by scope and an audit trail of consent and banner events, from the optional consent module.
- Site performance: Core Web Vitals (LCP, FID, CLS, INP, and TTFB) by page and device, from the optional Core Web Vitals module.
Who uses the Digital Analytics Data Model Pack?
For data and analytics teams
- Query the tables directly. Analysts can query views, sessions, users, and conversions with SQL, in a notebook, or from your BI tool.
- Start from modeled tables, not raw events. The packages process only new data on each run, with an optimized incremental method that lowers warehouse compute.
- Read and extend every line of SQL. The code is on GitHub. Build your own models on the same incremental logic instead of maintaining a separate pipeline.
- Set the definitions yourself. Change channel grouping, conversion definitions, and session and user identifiers to match how your business works.
- Keep bots out of your metrics. A bot filter applies across every model in the Unified Digital package, and you can tune it.
- Schedule the packages to run alongside your Snowplow pipeline. Teams that prefer can run them in their own dbt project.
For marketing, growth, and product teams
- Attribution you can explain. Compare four models side by side and see the logic behind every number, instead of trusting a black box.
- ROAS on your own data. Add spend data and measure return on ad spend by channel and campaign.
- Funnels and engagement in your BI tool. Analyze conversion, content, and retention without waiting for a new pipeline.
- One set of definitions. Every team reports from the same sessions, users, and conversions.
- Returning visitors stay recognized. Snowplow can recognize returning users for up to two years, even as browsers shorten cookie lifetimes.
Looking for a GA4 or Adobe Analytics alternative?
Moving off a packaged analytics tool usually means rebuilding the reports your teams depend on. The pack gives you sessions, users, conversions, channel reporting, and attribution as tables in your warehouse from your first model run, so the move doesn't start from raw events.
- Familiar channels from day one. The default channel grouping draws on GA4 source categories, and you can adjust it.
- Run both while you switch. The Snowplow web tracker can capture Google Analytics cookie values on every event, so you can join the two datasets during a parallel run.
- Plan for history. The pack models the data Snowplow collects, so teams typically keep their current tool running until Snowplow has enough history for year-over-year reporting.
How does the Digital Analytics Data Model Pack work?
- Snowplow trackers collect events from your website and apps.
- The Snowplow pipeline validates and enriches each event and loads it into your warehouse.
- The dbt packages then build views, sessions, users, conversions, attribution, and media tables, which your BI and AI tools query directly.
Supercell now joins Snowplow web data with its server-side game data on a shared ID, for one view of player behavior from website to game.
"Our previous digital analytics tool was very limited for us. We couldn't fetch the data easily; we couldn't send custom identifiers, and there were GDPR concerns. Most importantly, we couldn't integrate it with our other data sources to get a full picture of player behavior, from marketing websites to in-game actions."
Digital Analytics Data Model Pack FAQs
Does the Digital Analytics Data Model Pack include dashboards?
No, the pack provides data models rather than dashboards. Its dbt packages build tables in your warehouse, including views, sessions, users, conversions, and attribution. Your team connects its own BI tool, notebooks, or AI tools to those tables, so every report uses the same definitions and you aren't tied to one interface.
How is it different from exporting GA4 data to BigQuery?
The GA4 export gives you raw event tables that your team still has to model, while the pack ships modeled tables built from Snowplow data. You get sessions, users, conversions, channel reporting, and four attribution models in your warehouse, with SQL you can read and change. The data comes from your own Snowplow pipeline, so you decide what's collected and how it's defined.
Can I customize channel groupings and conversion definitions?
Yes, both are configurable in the dbt packages. The default channel grouping uses campaign parameters such as UTMs and GA4 source categories, and your team can override it with its own rules. A conversion can be any single event that matters to you, such as a sign-up or a visit to a specific page, and you can change the definition later and apply it to past data.
How does Snowplow define a session?
By default, a web session ends after 30 minutes of inactivity, and your team can change that timeout in the Snowplow tracker. The Unified Digital package builds its sessions table from those sessions and includes any session with at least one page view, screen view, or page ping. If your business needs a different definition, you can set custom session and user identifiers in the package.
Does it work for mobile apps as well as websites?
Yes, the Unified Digital package models web and mobile app events into the same views, sessions, and users tables, using page views on the web and screen views in apps. The Media Player package also covers video and audio in mobile apps and on Roku, so you can analyze websites and apps with one set of definitions.
Do I need to run dbt myself?
Not necessarily. Customers on Snowplow Customer Data Infrastructure can schedule the packages to run alongside their Snowplow pipeline, so the tables stay current without a separate scheduler. Teams that already use dbt can add the packages to their own dbt project instead and customize them there. Either way, the models run in your warehouse and process only new data on each run.
What is required to support attribution modeling within a data warehouse?
You need event-level data with campaign and referrer details, a clear definition of what counts as a conversion, and a way to connect each user's sessions over time. The Attribution package handles the modeling: it builds paths to conversion from the Unified Digital views and conversions tables, then credits channels and campaigns under four models. Add your ad spend data to calculate ROAS.