Orders from customers who won't do EDI: how to automate email and PDF orders

The orders that never come in as EDI

Here's the mix at a lot of 40 to 400 person manufacturers and distributors. Two or three big customers send EDI. Their orders land in the ERP with little or no typing. Everyone else emails a PO. Sometimes it's a PDF printed from their system. Sometimes it's a spreadsheet, or a few lines typed into the email itself. Someone on your team reads each one and keys it in.

Those emailed orders are often most of your customers, even if they aren't most of your revenue. And asking them to switch to EDI rarely works. This article covers what those orders cost, why the customers who send them won't change, how to automate them anyway, and where EDI is still the right tool.

What the benchmarks say it costs

The best public numbers come from APQC, a nonprofit benchmarking group. They come with a caveat: both reports we use here were sponsored by Esker, a company that sells order automation software. The data is APQC's, but keep the sponsor in mind.

In its 2016 report, Cutting the Costs of Sales Order Processing, APQC compared 608 organizations on the total cost to manage sales orders, per order placed. Bottom performers spent almost 8 times more per order than top performers: $40.87 against $5.11. That figure covers the whole order process, not just typing.

What drives the gap? APQC is direct about it: "the most significant cost driver is labor." That means the time to enter the data, start the work that fills the order, and update the customer. It also means the time lost when part of an order goes missing, when someone makes a data entry error, or when a change to the order arrives. Changes are the quiet cost. A customer who emails a PO also emails the revision, and someone has to find the first order and fix it.

The same report split costs by how orders arrive. APQC compared traditional channels, which it describes as paper-laden methods, with new, digital channels. It notes that companies that capture orders electronically, even orders received by email, spend less. The gap was the same at both ends of the scale.

Average cost per sales order, by channel

APQC benchmarking data, 2016. Traditional (paper-heavy) channels, 755 organizations; new (digital) channels, 758. 2016 dollars. Sponsored by Esker.

1. Top performersDigital channels$2.00
2. Top performersTraditional channels$7.00
3. MedianDigital channels$4.00
4. MedianTraditional channels$12.60
5. Bottom performersDigital channels$6.00
6. Bottom performersTraditional channels$21.00

Source: APQC, Cutting the Costs of Sales Order Processing, 2016, Figure 2 (sponsored by Esker).

Look at the pairs. Top performers paid $7.00 per order through traditional channels and $2.00 through digital ones. Bottom performers paid $21.00 and $6.00. Either way, the traditional order costs about 3.5 times as much. At the median, the gap is a bit smaller: $12.60 against $4.00.

APQC's 2018 follow-up, Transform Customer Service and Operations Through Order Automation, also sponsored by Esker, found a similar pattern from a different angle. Organizations that measure and reward getting order data right the first time had a median cost of $10.00 per traditional order and $3.10 per digital one. Those that didn't paid $16.50 and $5.10.

Two cautions. These are 2016 and 2018 dollars, and they come from large benchmark samples, not from companies your size. Use them to see the shape of the problem, then measure your own.

EDI has its own problems

EDI is not the error-free answer it's sometimes sold as. The 2018 APQC report says EDI can make order exceptions hard to process and hard to find. It also notes that up to a third of EDI orders present some kind of discrepancy. That figure comes from an Esker paper the report cites, so it's the sponsor's number, reported by APQC.

A discrepancy might be a price that doesn't match your contract, an old part number, or a ship-to you don't have on file. The EDI file arrives on time and in the right format. The data inside it can still be wrong. When that happens, the report says, the fix often pulls in both customer service and IT.

So the real goal isn't getting everyone on EDI. It's checking every order the same way, no matter how it arrived.

The goal isn't EDI for everyone. It's checking every order the same way.

Why many customers won't send EDI

EDI is a formal standard. In the U.S., most of it is X12, which describes itself as having more than 320 transaction standards. A purchase order is one of them, the 850. But each big buyer publishes its own rules for how to fill it in. Walmart's supplier site, for example, has a detailed guide to its 850 purchase order, segment by segment, including which fields are required.

That works when a huge buyer sets the rules and thousands of suppliers follow them. It works less well the other way around. When you're the supplier, you'd be asking your customer to set up EDI to send orders to you. For a 30-person contractor or a regional distributor, that means software, a setup project, and testing, all to place orders that they can already send in two minutes by email.

Most won't do it. Their buyer works in QuickBooks, a small ERP, or a spreadsheet. The PO prints to PDF, and email is how it goes out. Email is also simply how many buyers like to order. In a 2016 survey of more than 3,500 distributor customers, as reported by Distribution Strategy Group, about 74% said they order by email frequently or very frequently. The firm that wrote it up also promotes an order automation vendor in the same article, so treat it as one data point, and an older one.

Pushing customers to change is a sales risk too. If a competitor takes the PDF and you demand EDI, the customer may just order from the competitor.

How email and PDF order automation works

The idea is simple. Instead of making the customer change, you change what happens after the email arrives.

1. Watch the order inbox. Software picks up each new email and its attachments: PDF, spreadsheet, Word file, or plain text in the body.
2. Read the order. It pulls out the customer, PO number, ship-to, lines, quantities, prices and dates. Modern AI can read layouts it hasn't seen before, so you don't build a template for each customer.
3. Match it to your data. It finds the customer and ship-to in your ERP. It maps the customer's part numbers to yours, and it checks prices against your price list or the last order.
4. Flag what doesn't fit. A new ship-to, a price that's off, a part number that maps two ways, or a PO number you've already entered gets flagged, not guessed.
5. Draft the order for a person. Someone on your team sees the draft next to the original PDF and approves it. Then it goes into the ERP.
6. Handle changes the same way. When the revised PO arrives, software finds the original order and shows what changed.

That last step matters because changes are one of the labor costs APQC names. It's also where plain copy-and-paste tools tend to break.

The person stays in the loop on purpose. The software does the reading and typing. Your team still makes the calls that need judgment: price exceptions, credit holds, new customers and anything the software isn't sure about.

When EDI is the better answer

EDI isn't going away, and for some customers it's the right tool.

• When the customer requires it. Large retailers often make EDI part of the deal. SPS Commerce, an EDI vendor, says Walmart requires EDI from its suppliers, and lists the 850 order, 855 acknowledgment, 856 ship notice and 810 invoice. If you sell to a chain like that, you don't get a vote.
• When it's more than the order. Retail programs use EDI for ship notices and invoices too, with rules about what to send and when. Reading a PDF doesn't solve those.
• When volume is high and steady. A customer who sends hundreds of orders a week in the same format is worth an EDI connection. Once it's set up and tested, it's cheap to run.

For most suppliers, the answer is both. EDI for the few customers who require it or send heavy volume. Automated email and PDF intake for everyone else. Ideally, both feed the same checks, so an EDI order with a bad price gets flagged just like a PDF would.

What to measure before you start

You don't need a benchmark to know your own numbers. Pull a month of orders and count:

• How many arrived by email, by EDI, by portal and by phone.
• Minutes per order to enter, on a normal day.
• How many were changed after entry, and why.
• How many had an error found later: wrong part, wrong price, wrong ship-to.
• How many customers send fewer than five orders a month. These are the ones who will never set up EDI.

If emailed orders are a small share, a good template or a better order form may be enough. If they're most of your volume and most of your errors, that's where automation pays.

How Peekgentic does it

We build order intake around the ERP and the customers you already have. Software reads emailed POs from the order inbox in any layout, matches each line to your customers, items and prices, and drafts the order with the PDF beside it. Changes are matched to the original order. Anything that doesn't match is flagged for a person, who approves before it goes in. EDI orders can run through the same checks. See how an order-processing build works, or our guide to automating order entry.