Reference
Shopify and Xero, and
the decision it turns on
Connecting a storefront to a ledger looks like plumbing and is mostly modelling. The apps that do it are competent. Two businesses on the same app can get very different results because they answered one setup question differently, often without knowing it was a question.
That question is grain — whether each order arrives in the ledger as its own invoice, or whether a day of trading arrives as one summary journal. Much of what later feels wrong about the books traces back to it.
Four things have to cross, and only one of them is the sale
The other three are separate flows with separate timing, and they tend to surface at the first month end. An integration that carries only sales has moved the reconciliation rather than removed it.
-
1 — The sale
Revenue, tax and the customer, at whatever grain you chose. Every integration does this part, and it's the part integrations are judged on, which is why the other three surface late.
-
2 — The money, which arrives shaped differently
A card processor settles in batches, net of fees, on a lag, and a batch frequently spans a period end. So the bank shows one figure that matches no order, and something has to hold the relationship between the payout and the orders inside it. This is where revenue starts to look wrong rather than merely untidy.
-
3 — The fees
Taken out before the money arrives, which means they never appear as a payment and are invisible unless the integration books them explicitly. Left implicit, gross revenue and cost of sale are both understated by the same amount and margin looks better than it is.
-
4 — Refunds, and their timing
A refund moves money, revenue, tax and often a fee, rarely in the period the sale happened. Whether it lands against the original transaction or as a standalone credit decides whether a customer's history reads correctly a year later.
The grain decision, made once and lived with
There's no universal right answer, and there's usually an obvious one for a given business. What you can't do is change your mind cheaply later: the ledger fills up in whichever shape you chose.
One invoice per order
Every customer and every order exists in the ledger. You can answer a question about one buyer without leaving it, and credit control on trade accounts is possible.
The cost is volume. A few hundred orders a day becomes a few hundred ledger documents a day, reconciliation lists get long, and the accounts package starts being asked to be a CRM, a job it does slowly.
A daily summary journal
One entry a day carrying revenue, tax and fees by category. The ledger stays small, fast and readable, month end is quick, and the accountant is happy.
The cost is that customer-level detail lives only in the storefront. That's usually right for direct-to-consumer retail and wrong the moment there are trade accounts on terms, because you can't chase a debtor who isn't in the ledger.
The short version: summary if every sale is paid at checkout, invoices if anyone buys on terms. A business doing both usually needs both routes, and that's a deliberate setup, never a default.
What breaks after go-live
A second sales channel
The integration carries the storefront and knows nothing about a marketplace, a trade counter or a phone order. The ledger then reports a position that's confidently wrong on everything it can't see, and it doesn't flag the gap.
VAT that somebody else collected
Some VAT is collected by a marketplace rather than the merchant, and treatment differs by destination. This is a specialist question for whoever does your VAT — the integration will faithfully post whatever it was configured to post, including the wrong thing.
Money taken before delivery
Gift cards, deposits and pre-orders are cash now for a sale later. Booked as revenue on receipt they overstate the month and understate the next one, and the correction is manual every period until the mapping changes.
Stock, which the ledger isn't for
An accounting package answers what stock was worth at a date that has passed; a stock system answers what can be sold in the next thirty seconds. Asking the ledger for availability is the mistake accounting and stock is about, and stock control software covers what should hold it instead.
The connectors, as their makers describe them
How each one posts to the ledger, which ledgers it supports, how it's priced and how it reaches Shopify, taken from each maker's own site in September 2026. The difference that matters most is in the first line of each: a summary per payout, or a transaction per sale. It's the grain decision from above, already built into the product.
A2X
Posts each Shopify payout as summaries that reconcile to the bank deposit.
- Ledgers
- Xero, QuickBooks Online or NetSuite.
- Pricing
- Shopify from US$29 a month.
- Shopify
- App on the Shopify App Store.
Amaka
Syncs sales, payments and cost of goods into the accounting software.
- Ledgers
- Xero and Sage among others.
- Pricing
- Free tier, then Premium from US$18 a month billed annually.
- Shopify
- Billed through Shopify for Shopify users.
Link My Books
Imports each payout and breaks out sales, refunds, fees, VAT and payment gateways.
- Ledgers
- Xero or QuickBooks Online.
- Pricing
- Priced by monthly orders and number of channels, with a 14-day free trial.
- Shopify
- App on the Shopify App Store.
Shopify integration by Xero
Syncs sales, payouts and fees, as a daily summary or per sale.
- Ledgers
- Xero only. Built by Xero.
- Pricing
- Free on all Xero plans.
- Shopify
- Added from the Xero App Store.
Synder
Syncs individual transactions across 30+ sales platforms, with reconciliation.
- Ledgers
- Xero, QuickBooks Online, and Sage Intacct on higher tiers.
- Pricing
- From US$52 a month billed yearly, for 500 synced transactions.
- Shopify
- Shopify is a listed integration.
Choosing the connector
Which connector handles payouts properly, how it behaves across currencies, and whether Xero, Sage or QuickBooks is the right ledger all follow from the decisions above.
Settle the grain, the fees and the relationship between a payout and its orders first, and the connector choice narrows to one or two.
The useful first read is short: what grain is the ledger being fed at, whether fees are booked explicitly, and whether anything holds the relationship between a payout and the orders inside it.
Start the systems reviewTell us what you need
What the systems are doing now, and what you need them to do. We'll tell you whether it's a platform problem or something cheaper.