What Should I Automate in a Trading Company?

Start with the manual work happening between your orders, suppliers, customers and ERP

We already have an ERP. Why are we still doing all of this in Excel?

It is one of the most common frustrations inside a growing trading company. The sales order is in the ERP. The purchase order is in the ERP. The supplier information exists. The customer information exists. The stock is supposed to be there. Yet someone still needs to open an Excel file to know what is actually outstanding.

Someone still chases the supplier. Someone still checks the remaining quantity. Someone still compares a packing list with an order. Someone still has to work out which customers will be affected if a shipment is late. And when management wants to know what is really happening, somebody has to rebuild the picture.

If this sounds familiar, the answer is usually not to replace your ERP. And it is not to start by asking where you can put AI. The best automation opportunities in a trading company are often found in the work happening around and between the systems you already use — where people spend time checking, copying, reconciling, chasing and rebuilding information instead of moving the business forward. That is where you should look first.

A quick clarification: throughout this article, by "trading company" we mean an import-export, sourcing, wholesale or distribution business buying and selling physical goods — not financial trading.

Why do trading companies accumulate so much manual work?

Trading companies have a particular operational problem. They sit in the middle. On one side are customers. On the other are suppliers and manufacturers. Between them sit purchase orders, sales orders, stock, production, pricing, delivery dates, shipping, documents, payments, and dozens of exceptions.

And the information does not arrive neatly. Your ERP contains one part of the picture. A supplier sends an Excel packing list. Another sends a PDF invoice. The factory confirms a revised quantity over email. A freight forwarder sends a shipping update. Someone changes a delivery date. A customer asks for another 300 pieces. Stock is sitting at a supplier but has already been paid for. Another quantity is dropshipped directly.

The ERP is important. But the real world does not live entirely inside the ERP. That creates one of the defining activities of a trading company: reconciliation. People constantly reconcile what the system says with what is actually happening.

This is not just anecdotal. The UK government's recent Business Data Use and Productivity research found that smaller businesses remain more reliant on manual data collection, including stocktaking and sales-data entry, and identified Microsoft Excel as the most common internal data-collection and analysis tool across the businesses studied — highlighting the risk that an error in one part of a spreadsheet can propagate through linked calculations.

The OECD identifies a similar structural problem: SMEs tend to lag larger companies most strongly when digitalisation requires the integration of business processes across systems such as ERP, CRM and supply-chain management.

So if your company has an ERP and still runs important parts of the operation through Excel, email and people chasing each other, you are far from alone. The question is what to do about it.

What should a trading company automate first?

Don't start with the most impressive technology. Start with the work that creates the most friction every time another order enters the business. For most trading companies, the following operational areas are worth examining first.

1. Purchase-order and supplier-confirmation follow-up

Ask your procurement team: how do you know whether every supplier has confirmed what you ordered? If the answer involves opening an email inbox, checking an Excel tracker or asking someone whether they followed up, there is probably an opportunity.

A purchase order starts simple. You send 1,000 pieces. Then reality begins — the supplier confirms 800, the remaining 200 will be later, the production date moves, a colour is unavailable, a shipment is split, a quantity changes. Someone needs to update the information. Someone needs to remember what is still outstanding. Someone needs to chase again.

Someone has to chase every PO to know what is happening.

The consequence is not only irritation. Procurement becomes follow-up work rather than exception management, consuming staff capacity while problems are discovered later than they should be. A better operational model is one where normal purchase orders move without human attention, supplier commitments are visible, changes are detected, missing confirmations are identified, and people are alerted only when intervention is genuinely required. The goal isn't to remove the buyer — it is to stop the buyer spending the day asking "did they reply?"

2. Backorder monitoring

Backorders look manageable when the business is small. Then volumes increase. One order is partially delivered. Another is waiting for production. A supplier has shipped 700 of 1,000 pieces. Another 150 are coming next week. A customer has already received part of the order. Now somebody needs to answer what's still missing, what has shipped, what has been received, and which customer is waiting.

I need a clear view of what is still missing.

Without that view, staff repeatedly check orders one by one and customers receive inconsistent updates. A trading company should be able to automatically identify what remains outstanding, what changed, which backorders are within expectation, and which ones require action. You do not want people monitoring every backorder — you want people managing the backorders that have become a problem.

3. Supplier delay to customer impact

This is one of the most commercially important automation opportunities in a trading business. Knowing that a shipment is late is only the beginning — the real question is which customers are now going to be affected. To answer that, somebody may have to trace supplier → purchase order → product → outstanding quantity → available stock → customer sales orders → promised delivery dates.

The supplier is late. Which customers are going to be hit?

When that information is fragmented, a simple supplier delay becomes a small investigation, and by the time the answer is reconstructed, the customer may already be expecting delivery. The documented consequences include expediting cost, missed deliveries, lost trust, cancellations and margin loss. The system should not only tell you a supplier is late — it should help answer which orders are affected, which customers have committed dates at risk, and what requires attention.

4. Stock reconciliation and supplier-held inventory

One of the most dangerous questions in a trading company is also one of the simplest: do we actually have the stock? The ERP says 1,000. But some stock is reserved, some may have shipped, some is sitting at the supplier, some is dropshipped, some has already been paid for, some is incoming. And somebody says "let me check."

The stock in the system is not the real stock. We have to verify before confirming anything.

When system stock and operational reality diverge, people reconcile manually before making commitments to customers — with consequences including wrong promises, inventory errors and working-capital leakage. Supplier-held and dropship stock adds another layer: in one validated STREVIO workflow, stock held outside the normal ERP logic had to be manually deducted after sales and shipments ("we sold it, but someone still has to update the stock manually"). Every manual deduction introduces another opportunity for error. The objective should be one reliable operational view of what is available, outstanding, allocated and at risk.

5. Excel-to-ERP data entry

Why did we invest in an ERP if people are still typing information into it from Excel? Usually it's not that the ERP is useless — it's that operational information arrives in formats and places the ERP does not automatically understand. An employee exports information, cleans it, changes the format, copies it, uploads it, corrects errors, and does it again tomorrow.

We already have the ERP, but people are still copying everything from Excel.

The documented impact includes labour cost, processing delays, reduced capacity and data-entry errors despite the company already having invested in a system. For a growing business, this gets worse with volume — 100 rows become 500, 500 become 2,000. The employee gets faster at Excel, the business hires someone else, but nothing fundamental has changed. Where information can be safely mapped, validated and transferred automatically, people should not be acting as the connection between two digital systems.

6. Supplier documents, invoices and packing lists

Trading companies run on documents: supplier invoices, commercial invoices, packing lists, delivery notes, purchase orders, certificates, shipment documents, stock files. A surprising amount of skilled employee time can disappear into opening documents and extracting information that exists in a predictable format.

Gartner's 2025 supplier-information research describes supplier onboarding and data collection as error-prone and manual, pointing to automated data entry, validation and maintenance as areas where technology can reduce that burden. The important point isn't "AI can read PDFs" — the business question is why a person is reading this document, and what they need to do with the information afterwards. The useful workflow reads the document, identifies the supplier and PO, extracts the quantity and references, compares them with existing information, flags discrepancies, updates the relevant workflow, and asks a person only when something doesn't match.

7. Customer order-status enquiries

A customer emails asking when order 4837 will arrive. In some businesses, customer service checks the ERP, then asks operations, operations asks procurement, procurement checks the supplier, someone checks the shipment, and the answer travels all the way back. One customer question has created work for four people, and the customer waits while everybody is busy finding information the company supposedly already owns.

This is not about automating customer relationships — customers often want to speak to a human being. The problem is forcing that human being to conduct an internal investigation before they can answer. IBM describes supply-chain visibility barriers as existing between applications, organisational silos, suppliers and customers, making it difficult to see order status, changes and associated documents in a usable way. The better model gives the person dealing with the customer the operational truth immediately — improving service while simultaneously reducing internal workload.

8. Operational and management reporting

Every time I want the numbers, someone has to build me a report.

A report usually starts innocently. Management wants to see open orders, late POs, backorders, stock, supplier delays, sales, margins, customer commitments. The ERP report doesn't show exactly what management wants, so someone exports the data to Excel. Then another source gets added, then a formula, then another tab. Eventually the report becomes a weekly business ritual, and the person producing it becomes indispensable because only they understand how the spreadsheet works.

Excel itself isn't the enemy — it's an extraordinary business tool. The problem begins when management visibility depends on somebody manually recreating information the company already generated yesterday. Operational reporting should increasingly become a by-product of the operation itself, not another operation.

9. Margin and landed-cost reconciliation

This can be one of the highest-value problems in the whole business. The ERP says the order was profitable. Then someone reconciles purchase cost, freight, supplier charges, discounts, stock adjustments, additional logistics and the final accounting figures. Suddenly the margin is not what management thought it was.

I thought we made money until we checked the real numbers.

The documented consequences include hidden margin leakage, pricing mistakes and cash impact. Not every aspect of profitability can or should be automated, but the gathering and reconciliation of the information should not require a forensic investigation every month. When order, purchasing, stock, freight and accounting information can be connected, management gains something more valuable than a faster spreadsheet: confidence in the number.

The goal is not to automate everything

A trading company should not attempt to remove humans from every workflow. There are decisions where human judgment is the whole point:

• Negotiating with a strategic factory
• Deciding whether to accept a partial shipment
• Making an unusual commitment to a major customer
• Changing pricing because of a commercial relationship
• Choosing whether to expedite an important order at additional cost
• Handling a quality dispute
• Deciding whether a supplier deserves another chance

Technology should make these decisions better informed. It should not pretend they don't require a person.

Automate the checking, gathering, moving and monitoring. Keep humans responsible for the decisions that genuinely require judgment.

A real trading-company example: the ERP was not the problem

Another common trading problem: someone has to chase every PO

Your ERP and your operational reality are not always the same thing

ERP systems are essential — they are designed to record and manage transactions: sales orders, purchase orders, inventory, invoices, payments, receipts and other core business records. The OECD specifically identifies ERP systems as tools for managing and integrating information flows across finance, accounting, sales, inventory and purchasing.

But operating a trading business requires another level of questions: what changed since yesterday, which supplier didn't confirm, what is late, which customer is affected, which stock can we really promise, which exception needs somebody today? Those questions often cross several transactions, systems and external parties at once.

That is why companies can have an ERP and still depend heavily on Excel, email and WhatsApp. It does not mean the ERP implementation failed — it means there is a gap between recording the business and running the business. And that gap is where a lot of operational capacity disappears.

How should a trading company decide what to automate first?

Do not try to automate all nine areas at once. Choose the workflow where the business case is strongest. Evaluate each candidate against five questions.

How often does it happen?

A small task performed 100 times a week can matter more than a painful task performed once a quarter.

How much human effort does it consume?

Count not only the person doing the task but everyone checking, following up, correcting and reporting around it.

Does the workload grow with order volume?

This one is critical. If every 20% increase in orders creates roughly 20% more manual checking, the process is limiting your ability to scale.

What happens when it goes wrong?

Does somebody lose ten minutes, or does a customer receive the wrong promise? Does the business pay expedited freight? Does margin disappear?

How predictable is the normal workflow?

Predictable work with identifiable exceptions is usually a much stronger candidate than a process where every case requires a completely different commercial judgment.

The best first automation is usually the repetitive operational problem where growing transaction volume is creating manual work faster than your existing team can absorb it.

From checking everything to managing exceptions

There is one change of mindset that matters enormously in a trading company. Many teams operate through checking — every purchase order, every shipment, every backorder, every stock figure, every supplier response, every report. But there may be nothing wrong with 95% of those transactions. People are spending their time checking the 95 to discover the five.

A better operating model is management by exception. The normal flow continues, and the business automatically watches for the things that matter: supplier hasn't confirmed, delivery date changed, quantity mismatch, backorder exceeded tolerance, stock position inconsistent, customer commitment at risk, invoice doesn't match the PO. Then the relevant person gets involved. Instead of employees acting as monitoring systems, the system monitors operations and employees apply judgment where it matters.

More orders should not automatically mean more people

A trading company grows — more customers, more suppliers, more orders, more shipments, more complexity. The question is whether operational workload has to increase at exactly the same rate. It shouldn't.

More purchase orders should not automatically mean proportionally more supplier chasing. More customer orders should not automatically mean proportionally more status checking. More shipments should not automatically mean proportionally more spreadsheets. More transactions should not automatically mean proportionally more reporting work.

Operational Capacity: the ability to handle more customers, orders and work without proportionally increasing people, cost or complexity.

The OECD's current SME research makes the broader point that digitalisation can improve SME competitiveness and operational efficiency, while process integration remains one of the areas where smaller companies face the greatest difficulty. STREVIO operates inside that gap — connecting the systems and information your trading company already uses, improving operational visibility, removing repetitive work, and making sure people spend more of their time dealing with customers, suppliers, decisions and exceptions rather than reconstructing information.