AI Bookkeeping for Ecommerce: Inventory and COGS Done Right

Ecommerce breaks generic bookkeeping in one specific place: the moment you buy inventory, spend it, and try to figure out what a sale actually cost you. If your books count a 20,000 dollar inventory purchase as an expense the day you pay for it, your March looks like a disaster and your April looks like a miracle, and neither number is true. AI bookkeeping for ecommerce has to understand that inventory is an asset until it sells, and that cost of goods sold gets recognized against the sale, not the purchase.

Why ecommerce books go wrong

Most small-business bookkeeping treats money out as an expense. For a service business that is roughly correct. For a product business it is a lie. You spend cash on stock that sits in a warehouse. That cash turned into an asset, not a cost. The cost hits your P&L only when a unit ships to a customer.

Get this wrong and every downstream number is garbage. Gross margin swings wildly. You cannot tell a good month from a lumpy purchasing schedule. You overpay or underpay taxes because your profit is fiction. And you make pricing decisions on a cost number that has no relationship to reality. This is the core of what I mean in what automated bookkeeping gets wrong: the defaults are built for the wrong kind of business.

What AI bookkeeping must handle for ecommerce

Three things separate an ecommerce-ready system from a generic one.

First, inventory as an asset with proper COGS timing. The system books stock purchases to inventory, then recognizes COGS as units sell, so your margin reflects what you actually earned per order.

Second, marketplace and processor fee decomposition. A single Shopify or Amazon payout is not one number. It is gross sales minus platform fees minus payment processing minus refunds minus ad spend deducted at source, all netted into one deposit. An ecommerce-aware tool splits that payout into its parts. A generic tool records the net deposit as revenue and quietly erases thousands in fees you paid.

Third, refunds and chargebacks as reversals, not new categories. When a customer returns a product, the sale reverses and the inventory ideally comes back. If your books treat a refund as a random expense, your revenue is overstated and your unit economics are wrong.

This is where an AI-native approach pulls ahead. Reading a messy Amazon settlement report line by line, matching each fee type, and mapping it correctly is exactly the pattern-heavy, high-volume work that machines do well and humans hate. It connects to how the system reasons about categorization generally, which I cover in how AI categorizes transactions.

The numbers an ecommerce owner needs to see

Once inventory and fees are handled correctly, the reports get useful.

• True gross margin per order and per SKU, after real landed cost and platform fees.
• Contribution margin after ad spend, because a product that sells only through paid traffic can look profitable and still lose money.
• Inventory value on hand, so you know how much cash is frozen in stock.
• Sell-through and dead stock, so you stop reordering products that do not move.

Ficary is built to keep inventory and COGS honest instead of forcing you to true it up manually at year-end with your accountant. You can see how it approaches the ecommerce case at ficary.com. The point is not fancy dashboards. The point is that gross margin means gross margin, every day, without a quarterly cleanup.

What to verify before you trust it

Ask how the tool handles landed cost. Freight, duties, and import fees are part of what a unit costs you, and a system that ignores them understates COGS. Ask whether it can decompose marketplace payouts automatically or expects you to journal the fees by hand. And ask how it treats inventory adjustments for shrinkage, damage, and samples, because those quietly eat margin.

If you sell across several channels, confirm the tool can reconcile each channel's payout against its sales report. A mismatch there is usually the first sign that fees or refunds are being dropped. This is also why cleaning your data before you move matters, which I walk through in clean your data before you migrate platforms.

Ecommerce is a margin business disguised as a revenue business. The founders who win know their real per-order profit cold. Your books cannot give you that number unless they treat inventory and COGS the way ecommerce actually works. Fix that first, and while you are at it, remember to own your customer list, not just your store, because the platform can change its fees on you any morning.