The annual exchange between a merchant and their accountant usually starts with a screenshot of a dashboard and ends several emails later with the accountant asking, again, for the raw data. That is not obstinacy: a dashboard shows a summary computed by rules nobody has documented, and an accountant needs the underlying rows so they can apply their own rules and defend the result to a tax authority. Four exports answer almost every question they will ask, and knowing which four turns a week of back and forth into one email with attachments.
- Orders, payouts, refunds and taxes, as raw exports rather than as summaries.
- Give a date range and say which timezone it is in, because that alone resolves most queries.
- They work on invoice logic and you work on operating logic, so the two sets of figures will differ legitimately.
- Your P&L is not their P&L, and neither is wrong; they answer to a tax authority and you answer to a Tuesday.
The four exports
Orders, at line level, with dates, amounts, discounts, taxes and shipping. Payouts, with the transaction list, because reconciling a payout line by line is where fees are itemised and where the bank movements are explained. Refunds, dated by when they were issued. And the tax reports, split by rate and by jurisdiction where you sell across borders.
Between them those four contain almost everything an accountant needs from the shop. What they do not contain is anything about your costs, which live in supplier invoices and contracts and go to the accountant from your own records rather than from the platform.
Say which timezone, always
A month boundary is where most reconciliation queries come from, and the fix takes one sentence in the email: state that the export is in the store's timezone and name it. Without that, orders near midnight on the last day of the period sit in either month depending on who is looking, and an afternoon disappears into a discrepancy that was never a discrepancy.
It is the same mechanism behind why your Shopify numbers never match internally, and it is more consequential in this direction because somebody is filing a return based on the answer.
| Export | Answers | Watch for |
|---|---|---|
| Orders, line level | what was sold and taxed | cancelled and test orders |
| Payouts with transactions | what reached the bank and why | fees, reserves, adjustments |
| Refunds | money returned and when | dated by issue, not by order |
| Tax reports | rates and jurisdictions | cross-border rules |
Why their numbers will differ from yours
An accountant works on invoice logic rather than operating logic: ledger dates, invoices, the tax calendar and a duty to be correct for a year. Your statement books costs to the days that caused them so you can steer a week. Those two produce different monthly figures from identical underlying data, and both are right for their purpose.
The mistake is cross-reading: steering ad budgets by figures that arrive with invoice dates two months later, or asking an accountant whether Tuesday made money when their system has no concept of Tuesday. Keep both, use each for its own job, and reconcile once a year rather than continuously.
The two things they will not do for you
They will not tell you whether a product is worth advertising, and they will not tell you what happened last week. Not because they cannot, but because the work is priced for compliance and delivered on a compliance cadence, and asking for weekly operating analysis from an annual accounting engagement is asking for a different service at the wrong price.
That is the boundary worth naming out loud, because a lot of merchants believe they already have management reporting when what they have is a correct year filed on time. Both are necessary and neither substitutes for the other.
The questions worth asking them
Since the exports are going anyway, three questions turn a compliance exchange into something useful. Whether your cost of goods is being recognised the way you think it is, whether stock is being valued on a basis that matches your own records, and whether anything about the structure of the business is costing money that a different arrangement would not.
None of those is a reporting question and all of them are worth more than the reporting. An accountant who is not spending their hours reconciling a spreadsheet has the time to answer them, which is the practical argument for sending clean exports in the first place.
Making the annual handover trivial
- Agree the four exports and the date convention once, then send the same package every period.
- Include a note listing anything unusual: a large cancelled order, a supplier credit, a change of gateway.
- Send your own cost records alongside, because the platform has none of them.
- Use a data export that produces real files rather than screenshots, so their tooling can read it.
- Keep a copy of exactly what you sent, so next year's questions can be answered against the same rows.
The relationship this buys
An accountant who receives clean, consistent, well-labelled exports asks fewer questions, charges for less reconciliation and has more attention left for the things that actually save money: the structure, the timing, the reliefs. The exports are the cheap part. The expensive part is the hour they spend working out which figure you meant, and that hour is entirely avoidable.