The order desk before the first wave
Most of your clients are straightforward. Their Shopify or Amazon orders flow into the WMS through the connector overnight. Then there is the homeware brand that emails a spreadsheet of wholesale orders for garden centres each morning, the client on a platform your WMS has no integration for, and the one whose ERP drops a CSV on an FTP server that someone has to check.
A person on the order desk works through these each morning. They copy lines across, look up SKUs, fix addresses that will fail at label creation, and hope they finish before the first pick wave is released. When they are off, orders get missed, and the client finds out when their customer chases.
Why the connectors never cover everyone
Your WMS vendor builds connectors for the channels most of their customers use. Your clients do not all sell through those channels, and B2B orders in particular come from purchasing systems, buyers' emails and spreadsheets.
| Source | The usual gap |
|---|---|
| Shopify, WooCommerce, Amazon | Connector exists but silently stops when a token expires |
| Smaller or regional platforms | No connector at all |
| Client ERP export on SFTP | File arrives, nobody is alerted |
| Wholesale orders by email | Free text or attached spreadsheet, keyed by hand |
| Marketplace orders held by the client | Forwarded in batches, often late |
Connectors that fail quietly are as much of a problem as missing ones. An expired login means no orders arrive, and no orders looks the same as a quiet day.
What the manual lane costs
Keyed orders have keying errors: a wrong SKU picked and shipped, a missing house number, a quantity of 10 instead of 1. Late orders miss the carrier collection and break the dispatch promise the client makes to their own customers. And the order desk becomes a single point of failure that caps how many clients you can take on.
The intake layer we build
- Collectors for each source: platform APIs where your WMS has no connector, SFTP folders, and dedicated email addresses for clients who send orders as attachments or text.
- Parsing of spreadsheets and email orders into a standard order structure, with each client's layout remembered.
- Validation: SKUs exist for that client, quantities are sensible, addresses are checked against postcode data before they fail at label printing.
- Order creation in your WMS through its API, with the source recorded on each order.
- A heartbeat check per client: if a client that normally sends orders every day has sent none by a set time, someone is told.
- An exception queue for orders that fail validation, showing exactly what is wrong.
Where the WMS connector works, we leave it alone. The layer covers what the connector does not, and watches the connectors that exist.
The order desk afterwards
Orders from awkward sources are in the WMS before the first wave without anyone typing them. The order desk deals with a short list of genuine problems, and a silent connector is noticed in the morning, not when a client rings in the afternoon.
Taking on a client with an unusual order source stops being a reason to hesitate.
Signs this is your order desk
- Someone keys orders from emails or spreadsheets every morning.
- A client's connector has stopped before and nobody noticed for a day.
- Address errors are found at label printing, not at order intake.
- You have turned down or delayed a client because of how they send orders.
- Only one or two people know how each client's orders get in.