Purchase Order Function in MMS: A Guide to Manual, Semi-Auto, and Auto POs

Every time a retailer decides to stock a product, that decision eventually becomes a single document that ties the business and its supplier together: the purchase order. In a modern Merchandise Management System (MMS), the purchase order function is where forecasting, inventory data, vendor agreements, and financial commitments all meet. Get it right, and stores stay stocked with the products customers actually want. Get it wrong, and you end up with warehouses full of unsold goods or empty shelves during peak demand. This is why understanding how purchase orders work, and the difference between manual, semi-automatic, and automatic ordering, matters for anyone studying retail technology.

What a purchase order really is 🔗

A purchase order (PO) is a formal document a retailer sends to a vendor to order goods. It is not just an informal request. It is a legal commitment to pay for the products listed on it, which makes it a binding agreement between the retailer and the supplier.

A well-structured PO captures several essential details: the product code, the quantity ordered, the cost price (calculated from the margins agreed with the vendor), and the total value of the order. Because every PO records exactly what was ordered, it also becomes the reference document against which deliveries are checked. The system can compare how much was ordered against how much was actually received, flagging any shortfall instantly.

Why fill rate becomes a key metric 🔗

This comparison leads directly to one of the most important supplier metrics in retail: the fill rate. Fill rate measures the quantity a vendor actually delivered against the quantity ordered, expressed as a percentage. According to retail supply chain benchmarks, a fill rate of 95% or above is considered best-in-class, while anything below 85% usually signals a problem.

Mature retailers track fill rate as a Key Performance Indicator (KPI) for each vendor. A consistently low fill rate is concrete evidence that a supplier is letting the retailer down, and it gives the buying team a data-backed basis to push for improvement or renegotiate terms. Without a system recording ordered-versus-received numbers, this conversation would be based on guesswork rather than facts.

The two-step release process 🔗

Best practice in retail does not allow a single person to both create and release a purchase order. Instead, the process is split into two steps to maintain control.

First, a merchandiser prepares the PO. This involves analysing recent sales, checking the current quantity on hand, and factoring in the vendor’s lead time, which is how long the supplier takes to deliver after receiving the order. Second, a buyer reviews this prepared order and approves it on the system.

The control point here is strict: the warehouse cannot receive any goods against a PO that has not been approved. This operational rule prevents unauthorised ordering and ensures that spending stays within budget. Many MMS platforms let businesses set workflow approvals and financial limits for each buyer, so the system itself enforces the discipline rather than relying purely on staff memory.

Manual purchase orders 🔗

A manual purchase order is one where the merchandiser does the number crunching themselves, using the MMS as a support tool rather than letting it decide. The person looks at sales history, current inventory, and the broader market situation, then uses their own judgement to set the order quantity.

The reason human judgement still matters is that systems struggle with context. Consider a sportswear retailer stocking team T-shirts during a cricket World Cup. Sales might be booming, and a system looking only at recent data would suggest ordering more. But if India loses a crucial match, future demand for those T-shirts can collapse overnight. A standard MMS cannot capture this kind of event-driven shift on its own. More advanced business intelligence solutions can help, but even then, a human often needs to make the final call. Manual POs are best suited to fashion, seasonal, and event-sensitive products where context outweighs raw numbers.

Semi-automatic purchase orders 🔗

A semi-automatic purchase order shifts the calculation to the system while keeping a human in charge of the final decision. Based on parameters that the retailer configures in advance, the software works out how much to order and prepares the PO automatically.

These configured parameters typically include: sales data for a chosen number of past days, the current inventory position, the supplier’s lead time, and any season control settings that adjust ordering for festive or seasonal peaks. The software does all the arithmetic and drafts the order, but it does not release it. A buyer still has to approve the PO before it goes to the vendor.

This approach works very well for Fast-Moving Consumer Goods (FMCG) categories, such as packaged foods, toiletries, and household staples, where demand is steady and largely driven by past sales patterns. The system handles the repetitive calculations, freeing the buyer to focus on reviewing rather than computing.

Automatic purchase orders 🔗

An automatic purchase order goes one step further. The software both raises and releases the PO without any human approval at all. When stock drops to a set level, the system orders more and sends it straight to the vendor.

This is only suitable for core SKUs, products that have no fashion or seasonal element and sell at a predictable, steady rate. A basic white or blue formal shirt that is part of a permanent range, or a staple like a soft drink, fits this category well. Demand is stable, so the risk of over-ordering is low.

Even so, the risk never fully disappears. A sudden change, such as a health concern affecting cola sales, can leave the retailer over-committed before anyone notices. Because of this, automatic ordering should be switched on only under strict conditions: sales data must be 100% accurate, and the stock recorded in the system must match the physical stock on the shelf with very high discipline. If system stock and actual stock drift apart, an automatic PO can order goods the retailer does not need, or fail to order goods it does. Reliable automatic ordering depends entirely on clean, trustworthy data.

The extra data that makes a PO powerful 🔗

A purchase order can carry far more than just product codes and quantities. The additional data points captured at the PO stage turn it into a planning tool, not just an order form.

Delivery and logistics instructions 🔗

A PO can specify a staggered delivery schedule, asking the vendor to phase deliveries over time rather than dumping the entire order at once. This avoids overloading the warehouse. It can also carry a direct-to-store delivery instruction, telling the vendor to ship goods straight to a particular store instead of routing everything through a central distribution centre. Many systems support these multi-drop and flexible delivery options as standard.

Special terms and PO expiry 🔗

Sometimes a specific order needs terms that differ from the master contract with the vendor, and these can be recorded on the individual PO. A particularly useful field is the PO expiry date. If the vendor fails to deliver by this deadline, the order is cancelled automatically by the system, protecting the retailer from receiving stock that has arrived too late to be useful.

Capturing the true cost 🔗

A PO should also record other costs such as insurance and freight. Adding these to the basic cost price gives the landed cost, the true cost of getting the product to the retailer’s warehouse. The landed cost includes the item price plus freight, insurance, duties, and handling. Without it, a retailer might assume a margin is healthy when extra logistics costs have quietly eaten into the profit. Capturing these figures at the PO stage means margins are calculated on real numbers.

Cancellation and allocation at the PO stage 🔗

Plans change, and a good MMS allows a retailer to cancel a purchase order either fully or partially, within the limits set by the master agreement with the vendor. This flexibility means an order can be trimmed if demand forecasts shift before delivery.

Advanced systems also allow merchandise allocation right at the PO stage. Instead of waiting for goods to arrive before deciding which store gets what, the retailer can split the order across stores while raising the PO. This allocation is driven by the past sales data of that SKU in each store, so high-selling locations receive more units.

The big advantage of allocating early is that the vendor can pack the goods store-wise for cross-docking at the distribution centre. Cross-docking means goods move through the DC without being stored; they are simply sorted and sent straight out to stores. This is a well-known distribution efficiency technique that cuts handling time and warehousing cost, getting products onto shelves faster.

What do you think? If you were managing a large retail chain, which product categories in your stores would you trust to an automatic ordering system, and which would you keep firmly under manual control? And how much discipline do you think a business needs in its day-to-day stock counting before it can safely let software order goods on its own?

References

1. https://www.lightspeedhq.com/blog/purchase-orders/
2. https://www.spscommerce.com/resources/guide-to-retail-supply-chain-metrics/
3. https://retail-assist.com/purchase-order-management-definition/
4. https://learning.sap.com/learning-journeys/discover-retail-functions-and-business-processes-in-sap-s-4hana-retail/setting-up-purchase-order-management_cea727c0-12fb-45f0-8eea-dfb3235e9111
5. https://www.brightpearl.com/blog/understanding-landed-costs-profit-margin-formula
6. https://www.klavena.com/blog/the-complete-guide-to-landed-cost-calculation-for-ecommerce/