What exactly do you sync between 1C and a marketplace?
Start with the list of entities, not with the technology. Every item on it has a direction, an owner and a frequency. If you skip this step, the integration turns into a pile of scripts that overwrite each other.
Most sellers end up with the same six flows. Which side is the master for each one is a business decision. The developer only implements it.
- Products and cards: usually 1C owns the article, cost and the link to the nomenclature, while the marketplace owns the card content such as photos and descriptions. The key is the link between the 1C item and the marketplace identifier.
- Prices: 1C calculates the base price, the marketplace may add promotions and discounts. Decide whether the price goes one way or whether you read back what is actually shown.
- Stock: goes from 1C to the marketplace for the schemes where you hold the goods yourself. For the marketplace warehouse the data flows the other way, for information.
- Orders: from the marketplace into 1C as a customer order or a sale, with a status that you then push back.
- Returns and cancellations: from the marketplace into 1C, with the goods going back to stock or to write-off depending on their condition.
- Money: payouts and commissions from the marketplace reports, for accounting reconciliation. This is the most overlooked flow and the one the accountant asks about first.
FBS, FBO and the other schemes: what changes for the integration
The scheme decides who holds the stock and therefore who you must keep in sync. The names and exact rules differ between Ozon and Wildberries, so check the current terms in the seller cabinet and the official documentation before you design anything.
In FBO the goods sit in the marketplace warehouse and the marketplace itself counts the stock. You do not push stock for these items. In 1C the movement looks like a shipment to the marketplace and later sales. In FBS (and rFBS on Ozon) you keep the goods, so you send current stock to the marketplace and handle every order yourself: confirm, pack, hand over. This is where the integration earns its money, and where most mistakes happen.
Many sellers mix the schemes: part of the range is shipped to the marketplace warehouse, part stays at home. Then the same item has two stock figures with different owners. Model them as separate balances from the start. A single number per item is the root of a typical oversell.
- FBO: stock is counted by the marketplace, 1C tracks shipments to the marketplace and reads sales for accounting and reporting.
- FBS or rFBS: 1C is the source of truth, stock is pushed to the marketplace, orders are pulled and processed on your side.
- Mixed: two balances per item, one per scheme, never summed into a single figure for the marketplace.
- Schemes with own delivery to the buyer: the order flow resembles FBS, but check the specifics of each marketplace in its docs.
Who is the source of truth for stock?
For the goods you hold yourself, 1C is the source of truth. The marketplace shows what you told it. This sounds obvious, until a manager edits stock in the cabinet by hand, and the next sync silently overwrites it.
Two practical rules help. First, send available stock, not the book balance: subtract what is reserved for orders you have already accepted but not yet shipped, and keep a small safety buffer for the items you sell on several channels. Second, forbid manual stock edits in the cabinet, or at least know that they exist and detect them.
If you sell on your own site and on two marketplaces from the same shelf, the question becomes harder: who reserves first? Pick one place that does the reservation, normally the layer that receives all orders, and let every channel read from it. Without that, you get oversells that nobody can explain afterwards.
- Available stock = balance in 1C minus reserves for accepted orders minus the safety buffer.
- One reservation point for all channels, not one per marketplace.
- Manual edits in the cabinet are treated as an incident, not as a feature.
- A full reconciliation of stock on a schedule, in addition to event-driven updates.
A ready module or your own intermediary service?
There are ready 1C modules and extensions for marketplaces, and services that sit between 1C and the cabinets. For a typical trading configuration and a normal range they are the right first step. They are cheaper to start and have already met the common edge cases. Check three things before you buy: does it support your exact 1C configuration and version, how fast does the vendor react when a marketplace changes its API, and can you see what was sent and why.
A custom service between 1C and the marketplaces makes sense when the rules are yours. For example: non-standard stock logic across several warehouses, bundles and kits, a mix of schemes, several legal entities and cabinets, or data that must pass through your own system before reaching 1C. In that design 1C talks to one stable interface of yours, and the service talks to the marketplaces. A marketplace API change then touches one place.
The technical details of how such a service talks to the marketplaces (tokens, limits, queues, retries) are in the article on Wildberries and Ozon API integration. The ways to connect 1C itself are in 1C integration options.
- Ready module: one marketplace or two, a standard 1C configuration, a plain range, a small team and no unusual rules.
- Ready module with caution: you need a feature the vendor does not have, or you cannot see the exchange log.
- Own service: several legal entities or cabinets, kits, mixed schemes, custom reservation, or the need to keep a full history of what was sent.
- Own service also fits when 1C is old or heavily modified and you do not want to touch it more than necessary.
Orders, returns and statuses: the part that breaks first
Stock is the visible part. Orders and returns are where the real problems hide. An order must arrive in 1C exactly once. The marketplace can repeat a notification, your request can time out after the other side has already accepted it, and a worker can restart in the middle of a batch. Without protection you get duplicate orders or lost ones.
The approach is the same as for any integration: store the marketplace order identifier, make the creation in 1C idempotent, and keep a queue with retries. The principles are covered in idempotency and retries in APIs and background jobs and queues in Go.
Returns need business rules, not only code. A returned item may be fit for sale, damaged or lost. Each case means a different movement in 1C and a different effect on stock. Agree on these rules with the warehouse and the accountant before development, otherwise the integration will faithfully reproduce a mess.
Why stock drifts, and how to catch it
Mismatches happen even in a good integration. Typical causes: a request that failed and was not retried, two updates that arrived in the wrong order, a manual correction on one side, a sale that happened in the window between two syncs, or a return posted in 1C with a delay.
Treat drift as normal and build for detection. Event updates keep the numbers fresh. A scheduled full reconciliation finds what the events missed. It compares what 1C thinks with what the marketplace reports, and produces a list of differences instead of silently fixing everything.
Decide in advance what the system does with a difference: correct automatically within a safe threshold, or flag for a person. Automatic correction of large gaps is dangerous, because the cause may be a real loss in the warehouse.
- Log every outgoing stock update with the item, the value, the time and the response.
- Run a scheduled reconciliation and keep its result as a report.
- Alert on a queue that grows, on repeated failures and on a long silence from one side.
- Show the status in a simple admin view: last successful sync, errors, items with a mismatch.
- After launch, someone must own the alerts. See support and monitoring after launch.
When you do not need a custom integration
Be honest with yourself first. If you sell a few dozen items on one marketplace, process a handful of orders per day and keep the stock in a spreadsheet, a dedicated integration is premature. The marketplace cabinet and manual upload will cost you less than the project.
You also do not need your own service if 1C is not your main system of record. When stock really lives in a warehouse program or a spreadsheet, fix that first. Synchronising two unreliable sources produces an unreliable result faster.
A ready module or a ready service covers a large share of cases. Build custom only when you can name the rule that the ready option cannot express.
When you need a contractor
It is worth hiring a developer when manual work already costs you orders: oversells, late shipments, hours of reconciliation every week. Another sign is that the ready module has been tried and could not express your stock or order rules.
A good contractor starts with questions, not with code: which flows are in scope, who owns each one, what the schemes are, how many cabinets and legal entities, what 1C version and configuration you run. They should be able to show how the exchange will be logged and how you will see an error. If the answers come before the questions, be careful.
If this matches your situation, see our integrations service. It is also worth reading about exchange between 1C, CRM and ERP via API.