Skip to main content
Qixas Group
Solutions
Services
Industries
About
Resources

The Sales Order Agent for Distributors: When Customer POs Arrive in Six Different Ways

Microsoft's Sales Order Agent can take repetitive emailed orders off your customer service team. How well it works depends on your item references, your review rules and an honest pilot.

Floyd Chan7 min readDynamics 365 Strategy
Model delivery truck and shipping cartons on a laptop keyboard, representing online customer orders

Your distribution business receives customer purchase orders through email, PDF attachments, Excel files, customer portals, EDI connections and phone calls.

Your customer service team spends the morning turning those requests into sales orders. This is what the current workflow looks like:

  • Someone opens an email, identifies the customer, checks the part numbers, confirms quantities, looks up pricing and enters everything into Microsoft Dynamics 365 Business Central.
  • Then another employee checks the order, because entering the wrong item or quantity can create a much more expensive problem downstream.

The process works. But it does not scale particularly well.

Microsoft's Sales Order Agent offers a way to automate parts of that work.

The order intake problem: email PDFs, customer portals, EDI and the phone

Distributors rarely receive orders through one consistent channel. A large customer may submit structured purchase orders through EDI (Electronic Data Interchange), another may email PDFs, and a smaller customer may simply type an order into the body of a message.

Each channel creates a different processing requirement.

ChannelWhat it takes to process
EDICan automate structured transactions when the integration is configured correctly.
Customer portalsMay require dedicated integrations or manual retrieval.
Phone ordersStill depend on someone capturing the customer's request.
Email ordersOften contain enough information to create a sales order, but still require employees to interpret and enter it.

Microsoft's Sales Order Agent is designed to assist with sales order intake through supported email-based workflows in Business Central. It can interpret customer requests, help identify the relevant customer and items, and prepare sales order information for review.

But it is not a replacement for every order intake channel. It does not process telephone conversations, for example. So start by measuring how many orders arrive through each channel. If most of your volume already arrives through reliable EDI connections, the Sales Order Agent may have limited impact. If employees spend hours processing repetitive emailed orders, the opportunity is more substantial.

What the agent reads cleanly and what defeats it

The Sales Order Agent works best when customers provide information that is straightforward and easy to interpret. For example:

  • Customer: ABC Industrial Supply
  • Item: Part 45820
  • Quantity: 40
  • Requested delivery: November 30

Real customer orders are not always that clean. A PDF may contain several delivery locations, or the customer may include a handwritten note. An email might say, “Send the same items as last month, but double the second line.” Orders like these require historical context and interpretation.

Other complications include poor-quality scanned documents, ambiguous item descriptions, multiple purchase orders in one attachment and contradictory instructions between an email and its attachment.

The agent may identify information from supported inputs, but you should not assume every document format or instruction will be interpreted correctly.

A successful pilot should include difficult orders: customer documents containing unusual descriptions, multiple shipping instructions and incomplete information. The goal is to determine which orders can be processed reliably and which should be reviewed manually.

Item matching against customer part numbers, which is where this lives or dies

This is where distributors should spend most of their preparation time.

Your customer calls a product one thing. Your supplier calls it another. Business Central uses an internal item number. A customer may order part AB-2400, while your inventory system identifies the same product as ITEM-00847, and another customer uses an entirely different reference for it.

If the relationship is documented correctly, the order can be matched. If it is not, the agent has to interpret the description or identify a possible match. That introduces risk.

Business Central supports item references, including customer-specific references, that can help connect external product identifiers to internal items. Before implementing the agent, review customer item references, units of measure, product descriptions and discontinued items.

Pay particular attention to packaging. A customer ordering ten cases is not necessarily ordering ten individual units. If your item master does not clearly distinguish cases, cartons and individual pieces, an incorrect match can affect inventory, pricing and fulfilment.

The same applies to substitute products. An agent should not assume that a similar description means two items are interchangeable. A replacement part may have different specifications, compatibility requirements or customer approvals.

Note: The Sales Order Agent is only as dependable as the product information and matching rules behind it.

Pricing, contract terms and credit holds: it will not decide for you

Creating a sales order is not the same as approving a commercial transaction.

Distributors often maintain customer-specific pricing, volume discounts, negotiated agreements and payment terms. Some customers have credit restrictions. Others require special freight arrangements or delivery commitments.

Business Central can calculate pricing using configured sales pricing and discount rules. It can also support credit-limit warnings and other controls. But those capabilities must be distinguished from the Sales Order Agent's responsibilities. The agent should not independently negotiate a price, approve an exception to a customer contract or authorize an order that violates company policy.

Suppose a customer submits a purchase order using a price from an expired agreement. The agent may extract the requested price, but that does not mean the price should be accepted. Or consider a customer requesting an unusually large order while carrying an overdue balance. The fact that the order can be entered does not mean it should be released.

Define which commercial decisions require employee review. For example, a requested price that differs from the configured Business Central price should trigger an exception rather than an automatic override. Credit-limit and credit-hold handling should follow your established controls, not assumptions about what the agent will enforce.

Setting what escalates to a human, by dollar value and by exception

Not every sales order needs the same level of review. A repeat order for standard products from an established customer may require minimal intervention. A high-value order containing unfamiliar items deserves more attention.

The practical approach is to establish an escalation policy before deploying the agent. You might define an initial pilot policy such as:

  • Orders below $2,500 follow the standard review process when all information matches.
  • Orders from $2,500 to $10,000 require an additional check before confirmation.
  • Orders above $10,000 require supervisor approval.
  • Any order with an unmatched item, disputed price, unusual delivery request or credit issue requires human review.

These are illustrative thresholds, not Microsoft defaults or universal recommendations. The correct values depend on your average order size, margins, customer relationships and financial exposure.

There is another distinction worth making. Your business can define these escalation rules, but that does not mean the Sales Order Agent natively supports every proposed rule. Some controls may need to be implemented through Business Central workflows, extensions or a separate review process.

Start conservatively. It is better to review too many orders during the pilot than to discover that automation has introduced expensive fulfilment errors.

The tuning curve: what the first three weeks actually look like

Do not expect perfect automation on the first day. A sensible implementation begins with a limited set of customers and a controlled selection of orders. This is the three-week pilot structure we recommend.

Week 1: Establish the baseline

  • Select a small group of customers whose order formats and item references are reasonably consistent.
  • Process their orders through the agent while employees verify the extracted information.
  • Record every correction.
  • Separate errors into categories such as customer identification, item matching, quantity, unit of measure, pricing and delivery instructions.

Note: Do not measure success only by whether an order was created. Measure whether it was correct.

Week 2: Correct the recurring problems

  • Review the first week's exceptions. You may discover that several customers use part numbers missing from Business Central, or that others use inconsistent units of measure or descriptions.
  • Correct the master data and repeat the tests.
  • Document the cases the agent cannot handle reliably.

Week 3: Expand carefully

  • Introduce additional customers or more complicated order formats.
  • Compare processing time, correction rates and employee involvement against the original baseline.

Note: Do not increase the scope simply because the pilot has reached its third week. Expand when the results justify it.

Measuring it against what order entry costs you today

The business case should be based on the actual cost of processing orders. Start with three questions:

  1. How many orders does your team process each month?
  2. How much employee time does each order require?
  3. How much time is spent correcting mistakes?

Take an illustrative distributor processing 2,000 emailed orders per month, with six minutes of manual entry and checking per order. That is 200 hours of work. At a labour cost of $35 per hour, the monthly processing cost is $7,000.

Now suppose a pilot reduces average handling time from six minutes to three minutes across the same order volume. That would release approximately 100 employee hours per month.

It does not automatically translate into $3,500 of cash savings. The employees may still be needed for exceptions, customer service and other responsibilities.

Note: A proper business case should measure net time saved after reviews and corrections, along with error rates and any changes in customer service performance. If the agent saves two minutes on straightforward orders but adds five minutes of review to complicated ones, you need to know the distribution of those order types.

Summary

Before implementing the Sales Order Agent, examine how customer orders enter your business today.

Our recommendation: start with emailed purchase orders, clean up customer item references and establish human review rules before expanding automation. Measure accuracy and net processing time, not simply the number of orders the agent creates.

Qixas Group helps distributors implement and optimize Microsoft Dynamics 365 Business Central, including the processes that connect sales, purchasing, inventory and fulfilment. We also help organizations evaluate Microsoft Copilot and AI capabilities without assuming every process needs automation.

About the Author

Floyd Chan

Floyd Chan

President & Founder, Qixas Group

Floyd founded Qixas in 2007 after years of watching ERP projects fail for the same preventable reasons — junior staffing, slipping timelines, ballooning budgets. A Microsoft MVP and winner of the 2018 Directions Excellence Award, he is known across the Dynamics community as “the NAV genius.” He has been involved in more than 400 implementations and still reviews the hard ones personally.

  • Microsoft MVP
  • Dynamics NAV & Business Central
  • ERP Upgrades
  • Project Rescue

Frequently asked questions

What does the Business Central Sales Order Agent do?

It assists with sales order intake through supported email-based workflows. It can interpret customer requests, help identify the relevant customer and items, and prepare sales order information for an employee to review. It does not process telephone orders or replace EDI and customer portal integrations.

Is the Sales Order Agent worth it for a distributor?

It depends on how orders arrive. If most volume comes through reliable EDI connections, the agent may have limited impact. If employees spend hours entering repetitive emailed orders, the opportunity is more substantial. Measure order volume by channel before deciding.

How does the Sales Order Agent match customer part numbers?

It depends on your master data. Business Central supports item references, including customer-specific references, that connect a customer's part number to your internal item. Where those references are missing, the agent has to interpret descriptions, which introduces risk, so clean up item references and units of measure before a pilot.

Will the Sales Order Agent apply customer pricing and credit limits?

Business Central calculates pricing from configured sales prices and discounts and can support credit-limit warnings, but the agent should not negotiate prices, approve contract exceptions or release orders that break policy. Define which commercial decisions require employee review and treat price differences as exceptions.

How long should a Sales Order Agent pilot take?

Plan for about three weeks: a baseline week with a few consistent customers where every correction is recorded, a week to fix recurring master-data problems and retest, and a week to expand carefully. Expand only when correction rates and net processing time justify it.

Thinking about automating order entry?

We will look at how your customer orders arrive today, how clean your item references are and where a Sales Order Agent pilot would actually save time.

Book a Discovery Session