Ecommerce data systems
Ecommerce data warehouse
If you open Shopify, Amazon, Google Ads, and Meta to check yesterday's numbers, a warehouse can bring those reports together.
Data sources
Why it breaks
Why the reports disagree
Sales, ad, inventory, and finance data often disagree across Amazon, Shopify, and the ad platforms. Someone has to reconcile the definitions before the reports can be reviewed.
The systems that often need to agree
- •Amazon Seller Central
- •Shopify stores
- •TikTok Shop
- •Google Ads, Meta Ads, Amazon Ads
- •QuickBooks or NetSuite
- •Inventory systems
A practical architecture
What the warehouse does
A warehouse stores source data in one place, then models it for sales, cost, ad, inventory, and profit reporting.
The stack
01 / Integrate
Data integration
Copies source data from Amazon, Shopify, and ad platforms into the warehouse.
Fivetran: Managed connectors. Check each required source, table, and region before choosing it.
Airbyte: Connector software with self-hosted and managed options. Your team still owns setup, testing, and source changes.
Daton: An ecommerce-focused connector option. Test it against the Amazon and Shopify tables your reports need.
02 / Model
Data warehouse
Stores source data and runs the SQL models behind the reports.
Snowflake: Separates storage and compute and connects to common ingestion and BI tools.
BigQuery: Google Cloud's managed warehouse and a natural tool to test when the rest of the data stack is on Google Cloud.
03 / Use
Reporting and BI
Where people open dashboards, filter results, and investigate a number.
Tableau: Desktop authoring and published dashboards for visual analysis.
Power BI: Reporting and semantic models in the Microsoft ecosystem.
Looker: Browser-based BI with a modeled layer written in LookML.
Metabase: Open-source BI with a browser interface for queries and dashboards.
Case study
Seven ecommerce domains in one dbt project
We organized commerce, finance, inventory, refunds, fulfillment, marketing, and creator data in Snowflake and dbt for Tableau.
7
operating domains modeled together
120
models declared in the generated dbt manifest
10,105
order-line records reconciled during one connector migration
Build in order
A first-build sequence
Start with source access and one set of reports. Reconcile those numbers before adding more sources.
01 / Connect
Connect the sources
Connect the commerce and advertising sources needed for the first reports. Set up access and test each load.
02 / Reconcile
Reconcile and model
Build reconciliation reports, turn source data into reporting models, and compare warehouse totals with each platform.
03 / Review
Build and review reports
Build the executive summary and channel reports, then revise them after review.
Start with fewer sources
Start with the systems needed for the first reports. Add another source only after those reports reconcile to platform totals.
Sources to plan for
Start here
- Shopify API
- Amazon Seller Central
- Amazon Ads API
- Google Ads
- Meta Ads
- QuickBooks or NetSuite
Add next
- Google Analytics
- Klaviyo
- ShipBob or another 3PL
- TikTok Ads
Check access early
- TikTok Shop: confirm the required endpoints and account access
- Walmart Marketplace: confirm the required endpoints and account access
Decision guides
How to choose the tools
Compare each option against your required sources, cloud, report types, and the team that will operate it.
Snowflake or BigQuery for ecommerce data
Both are cloud data warehouses. Compare them against your cloud, query patterns, security requirements, and operating skills.
Snowflake
Runs on AWS, Azure, and Google Cloud. Storage and compute are configured separately.
Consider it when: Your data spans clouds or your team already runs Snowflake
BigQuery
A managed warehouse on Google Cloud. Google Cloud services can feed or query it without adding another cloud.
Consider it when: Your data and technical team already use Google Cloud
Tableau, Power BI, Looker, or Metabase
Compare how each tool handles modeling, authoring, sharing, permissions, and the reports your team already uses.
Tableau
Desktop authoring and published dashboards for visual analysis. Test it with one of your harder reports.
Power BI
Reporting and semantic models in the Microsoft ecosystem. Start here if the team already works in Excel and Microsoft 365.
Looker
Browser-based BI with a modeled layer written in LookML. Check whether the team can own that model.
Metabase
Open-source BI with a browser interface for queries and dashboards. Test permissions, sharing, and the queries your team needs.
How to test: Rebuild one report that people already use. Compare the totals, authoring work, permissions, and publishing steps before choosing a tool.
Fivetran or Airbyte for ecommerce
Compare source coverage and decide who will run connector infrastructure, monitor loads, and respond when a source changes.
Fivetran
Fivetran runs managed connectors. Your team still needs to test source coverage, monitor loads, and own downstream models.
Airbyte
Airbyte offers self-hosted and managed options. With self-hosting, your team runs the infrastructure and tests connector updates.
Daton or Fivetran for Amazon sellers
Start with the Amazon Seller Central tables, regions, and refresh timing you need, then test each connector against that list.
Test Daton when:
- ✓Most required sources are Amazon Seller Central, Shopify, or ecommerce ad platforms
- ✓Its connector covers the tables, regions, and refresh timing you need
- ✓You don't need sources outside its ecommerce focus
Test Fivetran when:
- ✓You integrate data beyond ecommerce (Salesforce, NetSuite, custom databases)
- ✓You need managed connectors outside ecommerce
- ✓Your team wants the vendor to run connector infrastructure
Triple Whale or a custom data warehouse
Triple Whale provides packaged ecommerce reporting. A warehouse lets you combine source data and define your own models.
Triple Whale
Test it when: You want packaged Shopify reporting and its definitions cover the questions you need to answer.
Check first: Required connectors, metric definitions, exports, and the reports you need to change.
Custom warehouse
Test it when: You need to combine brands or sources and define metrics in your own models.
Plan for: Implementation, reconciliation, monitoring, and ongoing model changes.
What to ask an implementation partner
If your team doesn't have a data engineer, ask who will own connector setup, reconciliation, models, reports, and monitoring.
Connector source coverage
Ask the partner to show coverage for the Amazon Seller Central and Shopify tables you need, plus how they reconcile orders, refunds, fees, and ad spend.
Models and reconciliation
Connector setup is only the first step. Someone still has to define models, compare totals with each source, and build the reports.
What to check before hiring
- •Work with the ecommerce sources you use
- •Examples of how they define attribution and ROAS
- •A source list, sample model, or case study you can inspect
- •A clear scope, source list, owner, and acceptance checks
Keep reading
Guides
Comparisons and setup guides for connectors, warehouses, and BI tools.
Amazon seller fees and payout reconciliation
Find the SP-API sources for payouts, SKU margin, FBA storage fees, and reimbursements.
Read guide →Fivetran, Airbyte, and Daton
Compare connector coverage, operating work, and warehouse load for Fivetran, Airbyte, and Daton.
Read guide →Power BI, Tableau, Looker, and Metabase
Compare Tableau, Power BI, Looker, and Metabase for modeling, authoring, sharing, and Excel workflows.
Read guide →BigQuery, Snowflake, and Azure
Compare BigQuery, Snowflake, and Azure against cloud fit, query patterns, and operating work.
Read guide →Stack builder
Choose a stack to test
Answer a short set of questions about your sources, team, and current tools. The result is a stack to test, not a buying decision.
Which platforms do you sell on?
Select all that apply
FAQ
Common questions
Start with the mismatch
Bring the reports that disagree
Bring the Shopify, Amazon, ad, inventory, or finance reports that do not agree. We can trace the definitions and decide where to start.
Review your reporting problemAbout Greyfield Data
Ecommerce data engineering and reporting
We build data warehouses and reports for ecommerce brands selling on Amazon, Shopify, and TikTok Shop. Client work includes multi-brand portfolios, DTC brands, and Amazon sellers.