Email order processing: how to automate orders from your inbox
Email order processing is the work of turning a customer’s emailed order into a sales order in your ERP: finding it in the inbox, reading it, matching the customer and products, checking stock and price, and keying it in. Automating it means software does the reading and checking, and a person approves the finished order instead of typing it.
For many UK wholesalers and distributors, the inbox is still where a large share of orders land. One customer sends a PDF from their buying system, another attaches a spreadsheet, and a third types “same as last week but 20 of the blue ones” into a reply.
What does email order processing involve, end to end?
Watch someone process an emailed order and you’ll see roughly the same sequence every time:
- Spot the order. The shared inbox holds orders alongside queries, remittance advices, delivery complaints and newsletters.
- Open everything. The PO might be a PDF attachment, with delivery instructions in the email body and an extra line added in a follow-up message.
- Identify the customer. The sender may be a buyer you’ve never heard from, or part of a group with several accounts.
- Read the lines. Product codes (sometimes the customer’s own, not yours), quantities, units and prices.
- Check against the ERP. Does each code exist, is there stock, does the price on the PO match the price on the account?
- Check it isn’t a duplicate. Customers resend orders, and two people can pick up the same email.
- Key it in. Create the sales order with the customer’s PO number as the reference.
- Close the loop. File the email, and reply if something needs confirming.
Each step is small. The difficulty is that all eight depend on the attention of the same few people, and steps 3 to 6 are where the costly mistakes creep in: the wrong account, the wrong product, last year’s price. We looked at how to put a figure on that in the real cost of keying in emailed orders.
Four ways to handle orders that arrive by email
Which option suits you depends on how many orders you receive, how varied they are, and where the time and errors actually come from.
1. Manual entry from a shared inbox
This is where most businesses start, and at low volumes it’s perfectly reasonable. A skilled person who knows your customers will catch things software won’t, such as a buyer who always means the 25kg bag when they write “bag”.
It struggles as volume grows, during holidays, and whenever the people who know the customers are needed elsewhere. It also leaves no record of what was checked, only of what was keyed.
2. Templates and parsing rules
Some tools let you define, per customer, where each field sits on their PDF. When a handful of large customers send identical, machine-generated POs, this can work well and cost little.
It breaks when a customer changes their layout, and it can’t cope with orders typed into an email, scanned paper, or a new customer whose format hasn’t been set up. Someone has to maintain the templates, and the work grows with every customer you add.
3. Shared inbox and workflow tools
Shared inbox software lets the team assign emails, tag them and track their status. If your real problem is orders getting lost, handled twice or left waiting because nobody knew whose job they were, this is a cheap and quick improvement.
It doesn’t remove the keying or the checking. Someone still reads every order and types it into the ERP.
4. AI capture with human review
Here the system monitors the order inbox, reads the email body and every attachment together, and fills in a fixed structure: customer, PO number, delivery details and lines. Each field carries a confidence score, and anything that isn’t in the order is left blank rather than guessed. Ordinary code then checks the result against your ERP, and a person reviews and approves it.
This handles the variety that defeats templates, including scans, spreadsheets and orders typed into the message. It costs more to set up than options 2 and 3, and it still needs a person in the loop, at least to begin with.
One more option sits alongside these. If a few large customers can trade by EDI, that’s often the better route for them, and email automation handles everyone else. We compare the two in order automation vs EDI.
What should order inbox automation do?
Whichever tool you choose, a few behaviours separate order inbox automation that helps from automation that creates a new class of problem:
- Keep non-orders visible. Queries and remittances shouldn’t vanish because the system decided they weren’t orders. They should be set aside where someone can see them.
- Read the whole message. The body, every attachment and any earlier messages in the thread, treated as one order.
- Leave blanks rather than guess. A missing delivery date that someone notices is far cheaper than an invented one that nobody does.
- Show its uncertainty. Fields it isn’t sure of should be flagged, so the reviewer knows where to look.
- Check facts with code. Product codes, stock, prices, customer matching and duplicate POs are questions your ERP can answer definitively.
- Keep the original beside the result. The reviewer should see the extracted order next to the email it came from.
- Use the customer’s PO number as the reference. It’s the number they’ll quote when they ring up.
How do you automate order entry from email without disrupting the team?
The safest projects change one thing at a time and prove it on real orders before relying on it.
- Collect a week of real orders. Save every emailed order for one normal week, the messy ones included. This sample shapes everything that follows and becomes the test set.
- Give orders their own address. A dedicated mailbox such as orders@ makes monitoring straightforward. Customers who send to a person’s address can be handled with a forwarding rule while you gradually update them.
- Agree what a complete order looks like. Which fields must be present before an order can go into the ERP, and which can be filled in later?
- Run it alongside manual entry. For the first stretch, process orders both ways and compare. Differences show you where the system needs tuning and where your own process has quirks nobody had written down.
- Keep approval on. Have a person approve every order to begin with. Once you trust the results, you can decide whether some orders, such as web orders that are already paid, could go through automatically. Our guide to designing an order approval workflow covers how to make that call.
- Add later stages when the first is settled. Capture, review and sending the order into the ERP are the natural first phase. Warehouse and shipping can follow once that’s running smoothly, and on our platform they’re added on request with a separate quote.
What does it cost to get started?
On the Zestcode Automation platform, the Essentials plan is £295 a month ex VAT with 200 credits, one for each purchase order processed, and setup to connect your orders inbox and ERP is from £1,500 ex VAT, quoted for the systems involved. Essentials keeps a person reviewing every order, while Standard adds the rule-based automatic approval described in step 5, and the Growth plan waives the setup fee. Every plan includes hosting, backups and platform updates, on a 12-month minimum term. Capture, review and ERP sync already exist on the platform, so you’re paying to have them set up around your systems. The pricing page has the full breakdown.
Cheaper routes exist. If templates or a shared inbox tool solve your particular problem, start there. A platform like ours earns its cost when your orders are varied, your volumes are meaningful, and you want the checks to follow your own rules rather than a generic product’s defaults.
If you’d like to see how your own orders would come through, our order capture page shows what the reading step looks like, and a discovery call is a good place to talk through a sample of your orders.