How to Connect Shopify to ChatGPT (Native App Setup)

You can connect Shopify to ChatGPT with two plugins that do different jobs. Shopify’s own plugin runs the store: tag @Shopify in a chat to check orders, change products or launch a discount.

The Windsor.ai plugin is for analysis. It gives ChatGPT your Shopify orders, customers, products and refunds, next to your ad platforms, GA4 or CRM if you connect them. Windsor.ai reads Shopify and does not change your store.

For the Windsor.ai route, connect your store in Windsor.ai with an app you create in the Shopify Dev Dashboard, then add the Windsor.ai plugin in ChatGPT.

🚀 Connect your Shopify data to ChatGPT with the Windsor.ai plugin now. Get started for free → onboard.windsor.ai/app/shopify.

Once your store is connected, you can ask complex questions such as:

• Which products drive high gross sales but low net sales after discounts and returns?
• Which SKUs are on track to run out of stock in the next 2 weeks?
• How many new vs returning customers did you have last month?
• Can you write a win-back email for customers who haven’t ordered in 60 days but previously bought 2+ times?
• Which products are frequently bought together and should be bundled?

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Is there an official Shopify app for ChatGPT?

Yes. Shopify publishes its own Shopify plugin for ChatGPT, and other plugins such as Windsor.ai read Shopify data for analysis. These are the routes, and when each fits:

• Shopify plugin for ChatGPT: free, from Shopify. Install it, then tag @Shopify in a chat. Shopify lists building a store, adding and editing products, checking orders and stock, launching discounts, and questions such as bestsellers, top countries, traffic sources and checkouts that did not convert. It reads and changes your store, and you can switch between stores after approving each one. It fits store tasks and questions about Shopify alone.
• Windsor.ai plugin: reads 435 Shopify fields, including refunds, line items, stock by location, UTM fields from the customer’s last visit and Shopify Payments transactions, plus any other source you connect in Windsor.ai. It fits net sales, retention and channel questions, and Shopify revenue next to ad spend. It only reads Shopify; the connectors that can make changes are in the write actions list. The steps below cover this route.
• CSV export: export orders or customers from the Shopify admin and upload the file to a chat. It fits a one-off question about a fixed period. Orders, customers and products come as separate files, and the file does not update after the export.
• Google Sheets or BigQuery: send Shopify data to a sheet or a warehouse table on a schedule with Windsor.ai (Shopify to Google Sheets, Shopify to BigQuery). It fits shared reports that keep their own history.
• Your own MCP server: in ChatGPT, full support for custom MCP servers, including write tools, is only in Business, Enterprise and Edu workspaces, where they are added from Plugins without Developer mode. Pro accounts get read and fetch tools only. It fits teams that build and run their own server.

Getting Shopify data into ChatGPT with the Windsor.ai plugin (2 steps, no code)

With the Windsor.ai plugin, connecting Shopify to ChatGPT takes two steps, with no code and no ongoing maintenance. The Dev Dashboard app takes a few steps in Shopify, and after that connecting ChatGPT takes a few minutes.

Just connect your Shopify store and install the Windsor.ai plugin directly inside ChatGPT.

📖 Full documentation: windsor.ai/documentation/windsor-mcp/how-to-integrate-data-into-chatgpt/.

What you need

• A Shopify store, and an app you create for it in the Shopify Dev Dashboard (it gives you the Client ID and Client Secret). Sign in at dev.shopify.com with the Shopify account you use for the store.
• A Windsor.ai account. The free forever plan includes 1 user, 1 data source and 1 account, with the last 30 days of data. Paid plans start at $19/month for 3 sources (Basic) and $99/month for 7 (Standard), billed annually (pricing).
• A ChatGPT account. We used ChatGPT’s Free plan for the steps and the example on this page. OpenAI says plugin availability depends on your plan, region, workspace and the ChatGPT interface you use, and in a Business, Enterprise or Edu workspace an admin can turn plugins on or off.

Step 1. Connect Shopify to Windsor

Go to onboard.windsor.ai/app/shopify and select Shopify as your data source.

Click Connect Shopify Store and enter your shop name, Client ID and Client Secret. You get the Client ID and Client Secret by creating an app for your store in the Shopify Dev Dashboard:

1. In the Dev Dashboard, create an app with the App URL https://onboard.windsor.ai, untick Embed app in Shopify admin, and paste the nine read scopes listed in the guide.
2. Choose Custom distribution and enter your store’s myshopify.com address to get an install link.
3. If you need orders older than 60 days, request the read_all_orders scope in Shopify Partners, then add it to a new app version.
4. Open the install link and install the app on your store.
5. Copy the Client ID and Client secret from the app’s Overview page.

After you click Connect Shopify Store in Windsor.ai, tick your store under Select Shopify Accounts.

⚙️ For detailed setup instructions with screenshots, check this step-by-step guide.

Managing multiple stores? Connect each one separately; they’re all queryable in a single ChatGPT conversation. Tracking more than one account needs a paid plan. The free forever plan includes 1 account.

Step 2. Install the Windsor.ai plugin in ChatGPT

Open Windsor’s plugin page in ChatGPT and click Connect, or Install plugin if that is the button you see. Grant access and you’re all set. You can also click Plugins in the ChatGPT sidebar and search for Windsor.ai. To disconnect later, open Settings, then Plugins.

To confirm everything is working, open a new chat, click + and choose Windsor.ai (or type @Windsor), and run this quick check:

What Shopify stores are connected to my Windsor account?
Show me the total revenue for the last 30 days.

Data confirmed? Go deeper with the advanced prompts further down.

ChatGPT requests the data from Windsor.ai when you ask. You do not need to refresh or export anything.

Example: low-stock products in a Shopify store in ChatGPT

We asked ChatGPT, through the Windsor.ai plugin:

Using the last 30 days of data, which product variants with status ACTIVE in my Shopify store windsor-guide-test have 15 or fewer units available for sale? List product title, variant title, SKU, price and units available, lowest stock first. Use product data only and do not show any customer data.

ChatGPT listed 15 active variants with 15 or fewer units in the windsor-guide-test store. Six have no units left: the four Gift Card values, The Out of Stock Snowboard and The Inventory Not Tracked Snowboard, whose zero means Shopify does not track its stock. The other nine have 9 or 10 units, five of them variants of The Complete Snowboard.

This question uses current stock levels, which the free forever plan covers. To estimate days until stockout, ask ChatGPT to compare these numbers with line_item__quantity from the last 30 days of orders.

Shopify prompts for deeper insights

These ready-to-use prompts answer questions that combine your orders, customers and products in one answer.

Just copy and paste. ChatGPT will pull the right data automatically.

💡 Good to know: If your Shopify app has the read_all_orders scope (see the Shopify Configuration Guide), Windsor pulls your full order history instead of the last 60 days. That means ChatGPT can compare this Black Friday to three years ago, calculate LTV across customer cohorts, and spot long-term trends. Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Revenue vs. reality: gross sales and net sales

Shopify Analytics reports gross sales and net sales (gross sales minus discounts and returns). The gap between them matters most for stores running frequent promotions or selling categories with high return rates.

ChatGPT can break this gap down by month, product and discount code and show which products keep the most net sales.

Calculate net revenue vs. gross revenue by month

Show total gross revenue vs. net revenue (after discounts and refunds) for each of the last 6 months.

How wide is the gap? Is it growing or shrinking?

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Find which products lose the most to discounts and refunds

For my top 20 products by revenue, show total discounts applied and total refunds received.

Which products have the biggest gap between gross and net revenue?

Measure the true cost of your discount strategy

How much total discount has been given in the last 90 days across all orders?

What % of gross revenue does that represent?

Which discount codes are responsible for the most discount spend?

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Customer intelligence: who’s worth attracting

Not all customers are equal.

A customer who buys once during a sale and never returns costs your acquisition budget and contributes very little long-term. A customer who orders four times a year at full price is your real business.

ChatGPT can separate the two in one prompt and tell you how to find more of the second type.

Identify your highest-LTV customers and what they bought first

Who are my top 10% of customers by customer total spent?

What was their first purchase: product, category, and order channel?

Is there a pattern that suggests what triggers high-LTV customer acquisition?

New vs. returning revenue split

For the last 90 days, split total revenue between new customers and returning customers.

What % of revenue comes from each group?

How does that compare to the previous 90 days?

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Find customers at risk of churning

Which customers have ordered 3 or more times but haven't purchased in the last 60 days?

List them with their total spend and last order date.

These are high-value customers worth a win-back campaign.

Write a win-back email

Based on customers who bought [product category] but haven't ordered in 60+ days, write a short win-back email.

Reference their previous purchase. Include a personalized reason to come back and a clear CTA.

Product performance: hidden gems, dead weight, and stockout risks

Shopify’s sales and inventory reports show sales by product, sell-through rate and days of inventory remaining. With orders, customers and stock in one chat, ChatGPT can also show which products appear in repeat customers’ orders, which products are bought together, and which SKUs could run out at their current sales rate.

Find your hidden gems: low volume, high loyalty

Which products have a relatively low total order count but appear frequently in repeat customer orders?

These might be undermarketed products that build long-term loyalty.

Spot products frequently bought together: bundle opportunities

Which pairs of products appear most often in the same order?

Are any of these combinations not currently offered as a bundle or kit?

Inventory velocity and stockout risk

For my top 15 products by units sold in the last 30 days, calculate the average daily sales rate.

Based on current inventory and that rate, which products will run out of stock within the next 14 days?

Identify dead stock

Which SKUs have had zero orders in the last 60 days but were selling regularly before that?

What changed: price or season?

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Discount codes and sales channels: do they bring quality customers?

Discount-driven customers often have a very different LTV profile from full-price buyers.

Understanding whether your promotional activity is growing the business or just inflating short-term order volume is one of the most useful analyses a Shopify store can run, and one that ChatGPT performs easily by joining order data with customer data.

Do discount code buyers come back?

Compare repeat purchase rates between customers whose first order used a discount code vs. customers who paid full price.

Do promo-acquired customers buy again at the same rate?

What's the 90-day LTV difference between the two groups?

Which sales channel produces the best customers?

Break down new customer acquisition by order sales channel for the last 6 months.

For each channel, what is the average customer total spent and customer orders count?

Which channel delivers the best long-term customer quality?

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Use ChatGPT as your store copywriter and strategist

Once ChatGPT understands your product catalogue, customer segments, and order patterns, you can ask it to do the work that flows from those insights: writing emails, drafting product descriptions, building promotional plans, and suggesting bundles.

Write product descriptions based on what customers buy

Identify my best-selling products. Analyze what makes them perform.

Then write 3 product description variations for low-performing products based on those patterns.

Build a promotional calendar from your sales data

Looking at my order data by month for the last 2 years, which months have the biggest revenue dips?

Suggest a promotional strategy for those slow periods based on my top-selling products and average order value.

Prompts that cover more than 30 days need a paid plan. The free forever plan includes the last 30 days.

Create a post-purchase upsell sequence

Write a 2-email post-purchase sequence for customers who just bought [product name], recommending the most logical next purchase.

Analyze Shopify data alongside your full business stack

Your Shopify store data shows what sold. But the bigger questions, like why it sold, which campaigns drove it, and which channels bring high-value customers, require data beyond Shopify.

🔗 Windsor.ai lets you connect 350+ data sources, so you can combine Shopify with Meta Ads, CRM, email platforms, and more for a complete cross-channel view.

Here are the most valuable combinations for Shopify store owners. Combining sources needs a paid plan. The free forever plan includes 1 data source.

• Shopify + Google Ads: Google Ads reports conversions via pixel. Shopify records actual paid orders. Connect both and ask ChatGPT: Which Google campaigns produce customers who buy again within 90 days? Where is Google Ads conversion value overstated because of return rates on specific products?
• Shopify + Meta Ads: Meta’s reported ROAS and Shopify’s actual revenue rarely match exactly. Connect both and ask ChatGPT to calculate true ROAS, actual Shopify net revenue from orders whose UTM source is Meta, divided by Meta spend, for each campaign (matched on the UTM campaign tag). Which Meta campaigns look efficient in Ads Manager but drive high-return products?
• Shopify + TikTok Ads: For DTC brands scaling on TikTok, the critical questions are CPA and customer quality. Connect both and ask ChatGPT: Do customers acquired through TikTok Ads have the same repeat purchase rate as customers from other channels? Which TikTok campaigns, by creative or audience, drive the highest 90-day LTV?
• Shopify + Klaviyo or Mailchimp: Connect your Shopify order data alongside email platform data and ask ChatGPT, by week: How do the revenue and conversions Klaviyo or Mailchimp reports for each flow or campaign compare with Shopify net sales from returning customers? Do weeks with more email sends also have more repeat orders?
• Shopify + GA4: GA4 captures site behaviour; Shopify captures purchases. Connect both and ask ChatGPT: For each session source, how do GA4 sessions and purchases compare with Shopify orders and net sales by last-visit UTM source? Which products have many views in GA4 but few orders in Shopify, a checkout friction signal?

What data Windsor.ai pulls from Shopify, and which fields matter most

Windsor connects to your Shopify store and gives ChatGPT access to detailed order, customer, and product data, covering 435 fields: 122 metrics and 313 dimensions. View the full list here.

Here are the fields that drive the most useful analysis.

Order data: the revenue layer

• order_total_price: Order value after returns, including taxes and discounts
• order_current_total_price: A duplicate of order_total_price, with the same current order value after returns
• order_net_sales: Gross sales minus discounts and returns, the figure Shopify reports as net sales
• order_total_discounts: Total discount applied per order, for measuring what discounts take from net sales
• order_current_total_tax: Tax collected, useful for net revenue calculations by market
• order_financial_status: Paid, refunded, partially refunded, pending, the payment state of each order
• order_cancel_reason: Why an order was cancelled, patterns here reveal fulfilment or product problems
• order_sales_channel: Where the order originated: Online Store, POS, Shop app, social, etc.
• order_tags: Custom tags applied to orders, useful if you tag by campaign, source, or customer tier
• order_created_at: Order date and time, the time dimension for all trend and cohort analysis

Customer data: the retention layer

• customer_total_spent: Total revenue from this customer across all orders, your LTV baseline
• customer_orders_count: Number of orders placed, distinguishes one-time buyers from repeat customers
• customer_is_returning: Boolean flag: has this customer ordered before? The single cleanest new vs. returning split
• customer_aov: Average order value for this customer, identifies high-value buyers vs. deal-hunters
• customer_tags: Custom tags on the customer profile, often used for VIP segments, wholesale, or acquisition source
• customer_created_at: When the customer was first acquired, for cohort analysis and LTV calculations
• customer_last_order_id: Their most recent order, calculate days-since-last-purchase for churn detection
• customer_state: Account status, enabled, disabled, invited

Line item data: the product layer

• order_line_items: Individual products within each order, the basis for product-level analysis
• line_item__quantity: Units sold per product per order, for volume and inventory trend analysis
• line_item__unfulfilled_quantity: Units not yet shipped, tracks fulfilment backlog and potential stockouts
• sku: Stock-keeping unit identifier, connects order data to inventory management

If something does not work

• Shopify shows “Couldn’t create app. Contains invalid scopes: read_all_orders”: create the app with the nine scopes first, request read_all_orders in Shopify Partners, and add it in a new app version once Shopify grants it (Shopify Configuration Guide, step 3).
• An error page opens in the Shopify admin after the install: Embed app in Shopify admin was left ticked. The app is installed anyway. Copy the Client ID and Client secret and continue.
• ChatGPT only sees the last 60 days of orders: your app does not have the read_all_orders scope. On the free forever plan, Windsor.ai returns the last 30 days in any case.
• Location names or unit costs come back empty: location names need the read_locations scope, and unit costs need the “View product costs” permission in Shopify.
• ChatGPT answers without your store data: click + and choose Windsor.ai, or type @Windsor, in that chat, then ask again.
• Windsor.ai shows “Disabled by admin”: your ChatGPT workspace admin has turned the plugin off. Ask them to enable it.
• A total differs from Shopify admin: check the store time zone (shop_timezone), which total you compare (Shopify’s Total sales adds shipping, taxes and duties to net sales; compare net sales with order_net_sales), the refund date (Shopify’s sales reports count a refund on the day it was made), and the date range and sales channel in both.

Conclusion

These questions help a Shopify store grow:

• Which products are worth scaling?
• Which customers are worth keeping?
• Which channels bring buyers who come back?

You can answer them with Shopify data in ChatGPT through the Windsor.ai plugin.

Connect your data once, and turn insights into action, from net sales analysis and LTV to emails, bundles, and campaign ideas.

The Shopify to ChatGPT connect page has the short answers and a second example. For e-commerce questions across Shopify, ad platforms and GA4, see how to use Windsor.ai for e-commerce and Windsor.ai MCP for e-commerce.

🚀 Connect your Shopify data to ChatGPT with the Windsor.ai plugin and start making smarter decisions today. Start Free Now.