AI agent for SHOPLINE: orders, attribution and ad spend together
SHOPLINE has the thinnest third-party tooling of any platform here, which makes a native connection disproportionately useful. It also has its own reporting API with its own definitions, and an attribution capture sparse enough that any answer about which channel drove sales has to say which source it used.
Connecting it, and what it asks for
- 1Open Integrations and click Connect on the SHOPLINE card.
- 2Authorize your International store (
your-store.myshopline.com) — two clicks, no token to create. - 3Ask for the sales summary or the change you want.
What the agent does with SHOPLINE
- Reads orders, products, customers and fulfilment state.
- Uses SHOPLINE's own sales report API, not just aggregated order rows.
- Derives channel attribution from ad click ids on the order, and reports the coverage alongside the answer.
- Joins store revenue to Google Ads, Meta and TikTok spend in one task.
- Raises daily threshold alerts on orders, revenue and AOV.
- Drafts store writes that pause for approval.
- Creates and retires discount codes, adjusts inventory (every pair is read first), edits product copy and tags, reprices variants and records marketing events for your campaigns — each checking the store's granted scope first and pausing for approval; deletions, absolute stock sets, repricing, deactivating a code (by status or by date), a discount of 50 % off or more and status flips wait for a person.
Two definitions you must not mix
In SHOPLINE's own sales report, net_sales is gross sales plus refunds and discounts (both negative), while total_sales adds shipping and tax. They are different numbers answering different questions, and quoting one while labelling it the other is a real reporting error rather than a rounding difference.
The agent names which figure it used. A report that does not is not safe to act on.
Attribution coverage, measured
SHOPLINE's own UTM capture covered roughly 11% of orders in live measurement, against about 85% when attribution is derived from the ad click ids that survive on the order itself. Both numbers are available, and every result carries its coverage so you can compare before choosing which to quote.
That gap is the reason this is stated rather than silently resolved: an 11%-coverage answer and an 85%-coverage answer can point in different directions, and only one of them should be used to move budget.
Frequently asked questions
- Does Agent Planners support SHOPLINE?
- Yes — SHOPLINE International connects natively, covering orders, products, customers, fulfilment and SHOPLINE's own sales reporting API, with store writes pausing for approval.
- Why does the agent report attribution coverage?
- Because SHOPLINE's own UTM capture is sparse — around 11% of orders in live measurement, against roughly 85% when derived from ad click ids on the order. Those two can point in different directions, so the source and its coverage are stated with the answer.
- Which SHOPLINE sales figure should I use?
- Whichever you name explicitly. net_sales is gross sales plus refunds and discounts, both negative; total_sales adds shipping and tax. Mixing them across one report is a reporting error.
- Can it give me a SHOPLINE conversion rate?
- Not from SHOPLINE alone — there is no sessions data, so there is no denominator. Connect GA4 and the question becomes answerable.
- Can it change my store?
- Yes, approval-gated: discount codes, inventory, product copy and tags, variant prices and marketing events, each checked against the scopes your store's authorization grants (a pasted private-app token records none, so SHOPLINE's own refusal speaks). Deleting a code, setting stock absolutely, repricing, deactivating a code (by status or by date), a discount of 50 % off or more and flipping a product's status always wait for a person, and so does a write through the generic GraphQL or REST gateway that deletes, changes store credit, points or inventory levels, reprices in bulk or touches storefront code.