Running an online shop means keeping two sets of books: the one your shop displays, and the one that has to add up at month-end. Between them sits a gap that classic bookkeeping simply does not have — and that gap is why e-commerce accounting takes far more time than the revenue figure suggests.

This guide describes what the gap is made of, what the path from document to locked month-end close looks like in five steps, and which of those steps are worth automating. One thing up front: not all of them.

Why e-commerce bookkeeping is different

In classic bookkeeping a transaction happens in one place: there is an invoice, there is a bank transfer, and the two belong together. In online retail the same transaction happens in four places at once.

Data source	What it gives you	What it does not give you
Marketplace / shop	Order, item, tax rate, delivery country	When and how much money arrives
Payment provider	Payouts, fees, chargebacks	Which order is behind them
Invoicing tool	The document you are required to keep	The actual flow of money
Bank account	The amount that lands	Anything about what it consists of

The decisive part is the right-hand column: none of these sources knows the whole transaction. An Amazon payout of €4,812.60 is not a revenue entry — it is what is left of sales, minus commission, minus FBA fees, minus refunds, plus or minus a withheld reserve. Book it as a single line and your revenue is understated and your fees are not recorded at all.

Then there is volume. A shop with 800 orders a month easily produces 3,000 accounting-relevant records across all channels. That is not a problem you solve with care; it is one you solve with structure.

The five steps — and what goes wrong in each

1. Collect every document

Completeness is the precondition for everything that follows, and it is where most bookkeeping quietly fails. A missing document does not announce itself: the total is there, it simply has nothing behind it.

In practice that means setting up automatic collection per channel rather than relying on manual exports — and actually running the check "number of orders in the shop = number of documents in the books". The German GoBD rules require exactly this traceability; a month with 812 orders and 799 documents is an audit finding, even if nobody notices it.

2. Match payments to orders

Payment reconciliation is the real core of e-commerce bookkeeping. A payout has to be turned back into its parts: revenue, fee, refund, chargeback, reserve.

PayPal, Stripe, Klarna and Mollie each present their settlements differently, and each names the same things in its own way. How to separate them cleanly per provider is covered in the post on payment provider accounting; for Amazon specifically, the post on settlement reports explains how a settlement becomes journal entries.

Automation is strong here — amount, date, reference and currency are enough for the large majority of matches. The rest stays manual, and that is fine: a partial payment, one transfer covering two orders, a chargeback weeks after purchase are cases that need a decision, not a rule.

3. Get the VAT treatment right

One price, many tax rates: as soon as you sell into several EU countries, the rate follows the delivery country rather than the product. Clean OSS separation is where genuine automation separates from the appearance of it — the fundamentals are in the post on OSS and IOSS, the arithmetic in the VAT calculator.

For B2B sales inside the EU, VAT ID validation comes on top. Reverse charge without a valid check is not a formality — it is a tax liability that stays with the seller.

4. Export something your accountant can actually import

An export is only finished when it arrives without errors at the other end. The most common DATEV import failures are rarely about substance — they are format questions: account length, date format, posting keys, adviser number. The EXTF error checklist and the DATEV export validator catch the typical cases before the file reaches the firm.

At higher volumes there is a further question: whether each order is posted individually or as a summary entry. Both are permitted; what matters is that the individual evidence survives and stays linked.

5. Close the month and lock it

The GoBD monthly lock is the step most automations skip, because it visibly accomplishes nothing. It is still the most important one: without the lock, a set of books can be changed after the fact, and books that can be changed after the fact are, if challenged, not books.

What can be automated — and what cannot

Honestly, the line is not where software vendors like to draw it.

Genuinely automatable, because it is rule-based and repeatable:

Collecting documents from shops, marketplaces and payment providers
Matching payments that carry an unambiguous reference
Determining tax rate and tax country from stored rules
Assigning accounts by channel, fee type and country
The export itself, including format and field validation

Not automatable, because it is a judgement:

How to explain a difference that no rule covers
Whether a chargeback is a revenue reduction or a bad debt
How to treat vouchers, discounts and partial returns — see the post on returns and chargebacks
Assessing borderline cases in tax law

An automation that claims the second list as well is guessing. The more useful kind takes the first list off your hands completely and puts the second list in front of you — with a proposal, a reason, and the option to decide otherwise.

What the manual route costs

The sum is rarely calculated, because the effort is spread across the month. Add it up — collecting documents, reconciling, chasing differences, exporting, correcting — and you land at two to four working days a month, before a single question from your accountant has been answered. The accounting cost calculator turns that into a figure for your own business, and the e-commerce accounting score shows in about three minutes which of the five steps is actually the bottleneck.

How KudTax maps onto this

KudTax is built around exactly those five steps: documents and transactions are collected automatically from Amazon, Shopify, eBay, Etsy, Kaufland and WooCommerce as well as PayPal, Stripe, Klarna and Mollie; payments are matched to orders; tax rate and tax country are derived from stored rules; and a DATEV- or Xero-ready journal batch comes out the other side. Every step stays traceable, with a timestamp and a link back to the document.

What KudTax deliberately does not do is guess. Cases that no rule covers arrive as a proposal with a stated reason — a human decides.

Conclusion

E-commerce bookkeeping is not a diligence problem, it is a structural one. The five steps are the same at 50 orders a month as at 5,000; the only thing that changes is whether they can still be done by hand. Bringing the four data sources together early saves you the reconstruction at year-end — and reconstruction is the expensive part.

The pragmatic place to start is steps 1 and 2: establish document completeness and automate payment reconciliation. The other three get easier almost on their own.

Bookkeeping for Online Sellers: The Guide to Automating It