Shopify integration

An ad platform counts what it saw: a form submitted, a click on a phone number, sometimes a purchase, often an estimate modelled from a sample. Shopify knows how many orders were placed, for how much, refunds deducted.

The gap between the two is not a tracking mistake, it is the nature of the two figures. Seeing it written down changes the conversation: you stop optimising a modelled CPA and start looking at what the day actually brought in.

• Orders — the count, per day, cancellations excluded.
• Revenue — net of refunds, in the store currency.
• Average order value — computed, never declared: the ratio of the two lines above.
• Customers — new and returning, counted separately.

The connection goes through OAuth: Shopify shows you what the app asks for, you accept on their side, and no password passes through our screens. The scope covers reading orders and customers — nothing else.

SaveAds therefore cannot create, modify or refund an order, change a price, or publish a product. This is not an internal policy that could be relaxed: the permission granted does not allow it, and a write permission on Shopify requires a review we have not asked for.

Ad-platform tables carry a cost, clicks, impressions and campaigns. A store has none of those. Forcing it into the same schema would have meant writing “cost: 0” on every row — asserting a zero where the honest answer is “not applicable”, and letting the engines conclude “no wasted budget” about an object where the question does not arise.

Sales therefore live in their own schema, and meet spend only where the comparison means something: a period, a currency, a channel.