9 ecommerce data sources worth connecting to a warehouse

Here are nine source types an ecommerce warehouse may need: commerce platforms, marketplaces, subscriptions, lifecycle messaging, ad platforms, attribution and surveys, finance systems, fulfillment and returns, and files people maintain by hand.

You rarely need them all at once. Start with the sources behind your first reports, and add one when it answers a question you can't answer today. The list below isn't a ranking, and connector support changes, so check any tool against the tables you need.

On experience: Greyfield's published case study covers Shopify, Klaviyo, Triple Whale, and Redo in a live reporting stack, and the Amazon fees guide covers Seller Central data. Other tools named here are examples, not a claim of completed work.

IDs from different systems aren't shared keys, and each section below repeats these fields as text.

1. Commerce platform

Examples: Shopify, plus other storefront platforms.

• Question: What sold, to whom, at what price, and what was refunded?
• Useful keys: order ID, line item, customer ID, SKU.
• Caveat: Shopify limits default order access to 60 days, so check history before you plan a backfill.
• Connect when: Nearly always first.

2. Marketplaces

Examples: Amazon Seller Central.

• Question: What sold on the marketplace, and what did it cost after fees?
• Useful keys: marketplace order ID, ASIN and SKU.
• Caveat: Fees come from several sources, and buyer data needs approved roles, so don't expect to match marketplace buyers to your site's customers.
• Connect when: The channel matters to your margin or your plan.

3. Subscriptions

Examples: Recharge, Shopify Subscriptions.

• Question: Who's subscribed, who skipped, and who cancelled?
• Useful keys: subscription ID, charge ID, the Shopify customer and order IDs.
• Caveat: Current state can overwrite history. Recharge's 2021-11 API docs say processed charges more than 90 days old no longer appear in charge responses, so check your API version, and keep your own event history.
• Connect when: Subscriptions are a real share of revenue, and your commerce data can't explain skips.

4. Lifecycle messaging

Examples: Klaviyo.

• Question: Who received and clicked what, and when?
• Useful keys: email or phone, event ID, campaign or flow.
• Caveat: Keep Klaviyo's profile ID, and map identities when profiles split or other systems use different identifiers. Opens can include Apple preloads.
• Connect when: You want to study retention at the customer level, not just campaign totals.

5. Ad platforms

Examples: Google Ads, Meta, Amazon Ads.

• Question: What did we spend, by channel, campaign, and day?
• Useful keys: date, account, campaign, ad.
• Caveat: Spend is the reliable part. Conversions are each platform's own credit, and they can overlap, so don't add them up.
• Connect when: You need CAC or spend by channel, which is usually early.

6. Attribution tools and surveys

Examples: Triple Whale, plus post-purchase survey tools.

• Question: What does a model say drove the order? What did the buyer say?
• Useful keys: order ID.
• Caveat: A model's credit and a buyer's memory are both evidence, not proof of cause. Keep the source and model name on every row, since exports vary by tool.
• Connect when: You already use one, and want it next to your orders instead of in a separate tab.

7. ERP and finance

Examples: NetSuite, QuickBooks.

• Question: What did we book, what did things cost, and what do wholesale accounts owe?
• Useful keys: GL account, invoice, purchase order, SKU, period.
• Caveat: The numbers follow accounting rules, so they won't match sales reports by default. Finance teams often need a named metric for each view. Access can be harder here, so ask early.
• Connect when: You need true costs, wholesale sales, or a tie to the closed month.

8. Fulfillment and returns

Examples: ShipBob and other 3PLs, Redo and other returns tools.

• Question: What did it cost to ship, how fast did it go, and what came back?
• Useful keys: shipment ID, order ID, SKU, tracking number.
• Caveat: One order can have several shipments, so joins can repeat rows. Invoiced costs can land after the order.
• Connect when: You need contribution by order, or you want to study delivery and returns.

9. Files and spreadsheets

Examples: cost sheets, sales targets, product mappings, promo calendars.

• Question: What does the business know that no system holds?
• Useful keys: SKU, month, channel.
• Caveat: Files break quietly. Give each one an owner, a fixed format, and a check for missing rows.
• Connect when: A report needs it, which is often sooner than people expect. Plan to replace a file if it starts to matter a lot.

What isn't on this list

Warehouses and transformation tools. Snowflake, BigQuery, and dbt store or shape data. They aren't sources. Ingestion tools move data, but they don't create it, so a tool on your stack diagram isn't a new source.

Before you connect a new source

Ask four questions:

1. What decision does this change?
2. Which key joins it to what we already have?
3. How far back can we get its history?
4. Who owns it when it breaks?

If you can't answer the first one, wait. Every source adds history, matching, and rules to maintain.

Start small, and let reports pull in the next source. If you're deciding which sources come first, the ecommerce data warehouse page has a suggested order.