Shopify to BigQuery & Snowflake data pipelines

When does a Shopify brand need a data warehouse?

Shopify analytics, GA4 and each ad platform are fine on their own. The trouble starts when a question spans them: what was blended CAC last month for new customers, by country, net of refunds? Answering that means exporting from three or four tools into a spreadsheet, and the next person who does it will get a slightly different answer.

A warehouse fixes that by putting all sources in one place with one set of definitions. Signs you have reached that point:

• Finance and marketing report different revenue for the same month.
• Every analysis begins with a round of CSV exports.
• You want LTV, cohort or forecasting work that needs years of order history in a queryable form.
• App dashboards cannot be joined with each other.

How does data get out of Shopify efficiently?

For anything beyond small stores, the right tool is the Shopify Admin GraphQL API’s bulk operations. A bulk query runs asynchronously on Shopify’s side and, when complete, produces a JSONL file at a signed URL, with one line per object. That avoids paging through millions of orders request by request, and the file streams cleanly into a warehouse load.

After the initial backfill, daily loads pick up orders created or updated since the last run, so edits and refunds on older orders are not missed.

What about GA4 and ad spend?

GA4 has a native BigQuery export. The daily export writes raw, unsampled event data to events_YYYYMMDD tables; an optional streaming export writes the current day to events_intraday_YYYYMMDD. Two points matter for planning:

1. The export does not backfill. Data before you link the property is not exported.
2. Standard GA4 properties have a daily export limit of 1 million events.

Ad spend from Meta and Google Ads is loaded daily at campaign level and joined to Shopify orders by date and, where tracking allows, by campaign.

What do the modeled tables look like?

Raw API output is not something to build dashboards on. We model it into a handful of tables your team can read:

Table	Grain	Used for
orders	One row per order	Revenue, margin, discounts, refunds
customers	One row per customer	First order date, channel, lifetime totals
daily_spend	Day × platform × campaign	CAC, ROAS, payback
sessions	One row per GA4 session	Funnel and landing page analysis
daily_kpis	One row per day	The executive dashboard

Every column gets a written definition. “Revenue” means one thing everywhere, and the dashboard in Looker Studio, Power BI or Tableau reads from these tables only.

How do you make sure warehouse numbers match Shopify?

A warehouse that disagrees with Shopify loses trust within a week, so reconciliation is built into the pipeline rather than checked once at launch.

Each daily run compares order counts, gross sales, discounts and refunds for recent days against the same totals from Shopify. If a load is late, partial or out of line, the check fails and someone is notified before the dashboard is opened. The same pattern covers ad spend, where daily totals are compared with what each platform reports.

Definitions are where most mismatches come from, so they are written down: whether revenue includes tax and shipping, when a refund counts, and how a returning customer is defined.

Who is this not for?

If your reporting questions are answered by Shopify analytics today and nobody is joining sources by hand, a warehouse adds cost without benefit. It is also not a fit if you want a fully managed connector product; we build in your own cloud account so you own the data and the code.

When reporting has turned into a weekly export ritual and the numbers still disagree, a pipeline from Asabisoft is a sensible investment.