Shopify to Snowflake integration tools compared on cost, refunds and duplicates
Snowflake's own Shopify connector is still a Preview feature on metered compute, and Snowflake ignores the primary key you declare. The right tool loads refunds, lands money as exact decimals per currency, merges on the Shopify id and costs the same in a busy month as in a quiet one.
The live demo needs no card, and Starter is $49 a month.
Field mapping auto-plugged · tap a port to rewire
Plug a source port into
Transform on this cable
JSON in
JSON out
5 sample records ready
Vendor documentation read 20 November 2026
Which Shopify to Snowflake integration should you use?
If you want a vendor-supported connector inside your own Snowflake account and can accept Preview terms, pilot the Openflow Shopify connector and budget its always-on compute. If you want a fixed monthly price, refunds loaded beside orders and every batch merged on the Shopify id, use a flat priced connector that backfills with a bulk export. CSV exports suit one analysis, not a model finance trusts at month end.
For Shopify on its own, with every API limit that governs an integration, see Shopify integration tools. For every other source into the same warehouse, see Snowflake ETL tools, and for the connector's runtime bill, Snowflake Openflow pricing.
What the Snowflake and Shopify documentation say, and what a buyer assumes
We read the Openflow Shopify connector page and its cost page beside Shopify's rate limit documentation, and wrote down what each line costs a finance team. The last column is what we would do.
| What is at stake | What the docs say | Why it matters | What we would do |
|---|---|---|---|
| Snowflake side | The Openflow connector for Shopify is a Preview Feature, read 20 November 2026. It uses the Admin GraphQL API, Bulk Operations and Snowpipe Streaming. | A Preview connector can change before it is generally available, and a finance close should not depend on terms that may move. | Pilot it, do not close the books on it |
| Objects | Orders, products and variants, customers, collections, inventory items and fulfillment orders. Other objects need an override. | Refunds, transactions and payouts are what turns gross orders into cash. Check where each one lands before you build revenue on it. | List the objects your report needs |
| Incremental runs | Watermark on updatedAt, createdAt or processedAt, in that order, and at most 250 child records per connection. | An order with a very long line item list can be cut short on an incremental run, with no error. | Compare line counts for your largest orders |
| Deletes and schema | Deletes set a soft delete flag. Schema evolution is not supported. The connector is read only. | Every dashboard needs the flag filter, and a new Shopify field means a manual change on your side. | Decide how deletes should look first |
| Runtime cost | No price on the connector page. Snowflake deployments bill compute pool credits per instance hour, 5 minute minimum, and a control pool runs with no runtimes. | A small store pays an always-on floor for a sync that changes a few hundred rows a day. | Do the instance hour sums below |
| Shopify limits | Pagination stops at 25,000 objects. GraphQL cost is 100 points a second on Standard and 2,000 on Enterprise, and a single query is capped at 1,000 points. | A paged backfill of a large store stalls or throttles. Bulk operations exist for exactly this. | Backfill with a bulk export |
The connector facts are on Snowflake's Openflow Shopify connector page and the limits come from Shopify's API rate limits documentation. The point arithmetic is worked through on Shopify API rate limits.
What the Openflow runtime bills in instance hours, before any rows move
Snowflake's Openflow cost page says a control pool runs even with no runtimes, and its example of one small runtime running for an hour is 2 instance hours. The credit rate per instance hour is in Snowflake's service consumption table and depends on your contract, so the table counts hours, not dollars. A month is 730 hours.
| Setup | Instance hours a month | How we got it |
|---|---|---|
| Control pool only, no runtime running | 730 | The floor the cost page example implies |
| One small runtime, 1 hour a day | 760 | 730 + 30 |
| One small runtime, 4 hours a day | 852 | 730 + 122 |
| One small runtime, 8 hours a day | 973 | 730 + 243 |
| One small runtime, always on | 1,460 | 2 instance hours for each hour |
Whether a given connector can run only a few hours a day is a question for your Snowflake account team, and the 5 minute minimum applies to every start. Whatever the answer, the Snowflake warehouse that your own queries use is a separate bill, as the credit floor table on the QuickBooks page works out. An Adapters plan is a fixed monthly price on pricing, whatever the order count.
Shopify to Snowflake connectors and integration tools compared
Billing units rather than price tags, because most vendors here meter or quote and any figure we printed would be stale within a quarter. Where a tool is the wrong fit, the last column says so, including for us.
| Tool | Approach | Best for | Billing unit | Watch for |
|---|---|---|---|---|
| CSV exports and Shopify reports | Download order and product exports and COPY INTO Snowflake by hand | A one-off analysis or one closing snapshot | No tool cost, paid in staff hours every month | Exports are flat summaries and nobody owns the refresh |
| Snowflake Openflow connector | First-party connector over Bulk Operations and the Admin GraphQL API | Snowflake-first teams comfortable with Preview software and Openflow runtimes | Snowflake compute pool credits per instance hour, no connector price listed | Preview status, soft deletes, no schema evolution, always-on control pool |
| Fivetran | Managed ELT with a Shopify connector writing tables into Snowflake | Teams with a data engineer and many sources on the platform | Metered by monthly active rows, confirm on the vendor pricing page | Order edits touch old rows, so model your busiest month |
| Airbyte | Open source or cloud ELT with a Shopify source connector | Engineering teams comfortable running pipelines | Self-hosted or usage-based cloud, confirm units with the vendor | Self-hosting means you own retries and upgrades |
| Hevo Data | No-code pipelines from Shopify into Snowflake | Analysts who want setup without code at modest volume | Metered by events, confirm tier limits with the vendor | Event billing punishes full reloads of order tables |
| Workflow automation | Trigger based tools inserting one Snowflake row per Shopify event | A single alert or audit table, not a reporting model | Billed per task, confirm on the vendor page | No backfill, and order edits and refunds never update the first row |
| Adapters | Visual mapping of Shopify objects onto typed Snowflake tables, bulk backfill, then merges on the Shopify id | Merchants and finance teams that want exact money columns and a fixed monthly price | Flat monthly plans from $49, no per-order or per-row fee | Not a replacement for a full commerce analytics suite |
Vendor billing units change, so confirm each one on the vendor's own page before you model a month. The same tools are compared more broadly on the best data integration tools.
Six facts worth knowing before you choose
Preview
Status of the Openflow connector for Shopify on 20 November 2026.
docs.snowflake.com
25,000
Objects a paginated Shopify query can reach, so large backfills need a bulk operation.
shopify.dev
250
Child records per connection on an incremental Openflow run.
docs.snowflake.com
5 min
Minimum billed on Openflow compute in a Snowflake deployment.
Snowflake, Openflow cost
1,000 points
Maximum cost of a single Shopify GraphQL query. Above it the query is refused.
shopify.dev
0
PRIMARY KEY or UNIQUE constraints Snowflake enforces on a standard table. Duplicates load without an error.
Snowflake, constraints
Six ways a Shopify to Snowflake pipeline is wrong while the log is green
Finance finds these at month end, not in monitoring. Each one exits zero.
| Failure | Why it is silent | The check |
|---|---|---|
| Revenue too high after refunds | A refund is a separate object. If only orders load, gross totals never move and net sales read gross. | Load refunds, net them by order and tie one closed month to the Shopify sales report. |
| Two currencies in one column | Shopify returns shop currency and presentment currency for money. Summing the wrong one mixes currencies and nothing errors. | Keep the currency code next to every amount and total by currency. |
| Duplicate orders | Snowflake accepts the PRIMARY KEY declaration and ignores it. A retried batch appended with INSERT doubles the rows and exits zero. | Count rows per Shopify id after every backfill. Any count above 1 is a failed load. |
| Test orders in revenue | Orders placed on a test gateway carry a flag. Without a filter they count as sales. | Exclude flagged test orders and compare the order count with the Shopify admin. |
| Days that shift at midnight | Timestamps are UTC while the store reports in its own zone. Casting to a bare TIMESTAMP loses the zone and orders slide into the wrong day. | Store TIMESTAMP_TZ and compare the daily count on the first and last day of a month. |
| Deleted products still counted | Soft delete means the row stays. A catalog report that ignores the flag lists products that are gone. | Delete a test product and confirm how it appears in the table. |
Six steps that decide whether this project works
Step 1
List the objects you report on
Orders, refunds, transactions, line items, products, variants, customers and inventory cover most merchant models. Orders alone cannot reconcile to a payout, because fees and refunds live on other objects.
Step 2
Declare money columns first
Create amounts as NUMBER with a fixed scale, keep the currency code beside each amount and store created as TIMESTAMP_TZ. Keep unstructured fields in a VARIANT column so new Shopify fields never break a load.
Step 3
Backfill with a bulk export
One bulk operation per object type brings the full history without paging, and tie each month to the Shopify sales report before moving on. Pagination cannot reach past 25,000 objects.
Step 4
Merge on id, never append
Every load is a MERGE keyed on the Shopify id, because Snowflake will not reject a duplicate for you. That one rule makes retries safe and lets order edits update the original row.
Step 5
Pick the schedule the readers need
Each run wakes a warehouse and bills at least 60 seconds. A daily revenue report does not need a 15 minute sync, and an inventory view might.
Step 6
Reconcile after every close
Count and sum by month and currency in both systems. A mismatch is a failed sync even if the log is green.
With Adapters the type decisions in step 2 happen when you map the field, so an order total lands as NUMBER beside its currency and created keeps its zone from the first run, and every batch is merged on id. The mapping for this pair is on the Shopify to Snowflake connector page.
Five jobs where you should not pick us
A comparison that never says the competition wins is an advert. These are the cases where something else on this page is the right answer.
- You are a Snowflake-first team that wants a vendor-supported connector inside your own account and can accept Preview terms. The Openflow connector is the least moving parts.
- You already run Fivetran or Airbyte across thirty sources into Snowflake and Shopify is one more. A second tool for one source rarely pays back.
- You only need storefront reports. Shopify's own analytics answers that without a warehouse.
- You need every Shopify object, including ones no preset covers. A connector with an object override or a full-schema loader will cover more.
- You need a push on every order the second it is placed. Every scheduled route here, ours included, runs on an interval.
Four questions to ask any vendor on this list
Question 01
Do you MERGE or INSERT?
Ask how the tool writes each batch into Snowflake. Snowflake will not stop a duplicate, so a tool that appends and deduplicates later leaves a window where your totals are wrong.
Question 02
What do I pay at twice my order volume, and with nothing running?
Ask for the bill at your busiest month, at double it and at zero. Per row pricing moves with your sales and an always-on runtime has a floor. A flat plan does neither.
Question 03
How do refunds, edits and deletes arrive?
Ask for the behavior by object and the column that shows the change. A vendor that cannot answer has not handled it.
Question 04
How are amounts and currencies stored?
Ask whether money lands as exact decimals with a currency code, or as strings you must convert. The second is where mixed currency totals come from.
Related Shopify and Snowflake guides
Shopify integration tools
Every Shopify API limit and the wider vendor landscape.
Snowflake ETL tools
Every other source into the same warehouse, and how each load is billed.
Shopify to Snowflake connector
Open the preset, review the field mapping and set the schedule.
Snowflake Openflow pricing
The runtime floor of Snowflake's own connectors, in instance hours.
Shopify API rate limits
Points per second by plan, the REST bucket and THROTTLED fixes.
Shopify to BigQuery
The same store data on Google's warehouse, with the Preview transfer compared.
Stripe to Snowflake
The payments side of the same warehouse, with Data Pipeline priced per charge.
Questions buyers ask about Shopify to Snowflake integration
- How do I connect Shopify to Snowflake?
- There are three routes. Snowflake's Openflow Shopify connector reads the Admin GraphQL API into tables in your account and is a Preview feature. A third-party or Adapters connector does the same on its own schedule. Or you run Shopify bulk operations yourself and load the files through a stage.
- Does Snowflake have a native Shopify connector?
- Yes, in Preview. The Openflow connector for Shopify replicates orders, products and variants, customers, collections, inventory items and fulfillment orders using Bulk Operations. It is read only, does not support schema evolution and runs on Openflow compute that Snowflake bills as credits, so it is native but not free.
- How much does the Snowflake Openflow Shopify connector cost?
- The connector page lists no price. In a Snowflake deployment you pay compute pool credits per instance hour with a 5 minute minimum, and a control pool runs even with no runtimes. One small runtime running all month is about 1,460 instance hours, against 730 for an idle control pool alone.
- Why do my Shopify order totals not match Snowflake?
- Usually refunds, currency or test orders. A refund is its own object, so gross order totals stay unchanged unless refunds are loaded and netted. Money comes as shop currency and presentment currency pairs, and test orders carry a flag. Reconcile one closed month against the Shopify sales report.
- Can Shopify data be backfilled into Snowflake?
- Yes, through bulk operations, which export a whole object type as one file instead of paging. Paging stops at 25,000 objects per query, so a store with years of orders cannot be backfilled by pagination alone. Run the bulk export once, then switch to changes since the last updated timestamp.
- Are deleted Shopify records removed from Snowflake?
- Not by default. The Openflow connector sets a soft delete flag on delete-tracked objects and never removes the row, which keeps history but means every report must filter the flag. Whatever tool you choose, ask how a deleted product or order shows up in the table.
- What is the best Shopify to Snowflake integration for a small merchant team?
- The one whose cost does not grow with orders or an always-on runtime, that loads refunds and money in the right currency, merges on the Shopify id so reruns never duplicate rows and tells you how deletes arrive. Run any trial on one closed month tied to the Shopify sales report.
For pushing warehouse results back into your storefront or CRM, see reverse ETL tools.
Land Shopify in Snowflake at a price that ignores your order count
Map each object once, declare exact decimals per currency, backfill in bulk, then merge only what changed on the schedule you pick. Flat plans from $49 a month, no overage fees.
The live demo needs no card, and Starter is $49 a month.