Manual picking is not an efficiency problem. It is a margin problem. With structurally narrow margins in wholesale trade in Portugal, every picking error, every duplicated route in the warehouse, every hand count that diverges from the system eats directly into the bottom line. Most companies that contact us already know they have a problem. What they do not know is that the cost of that problem is booked under returns, rework and customer credits — not under a line called "warehouse inefficiency". This article compares manual picking with integrated warehouse management and gives you a decision framework to use straight away in your next operations meeting. The premise is simple: keeping manual picking for longer than the business requires is not a conservative option — it is a decision to destroy margin, with an expiry date.
What you need before starting
Before comparing approaches, check whether you have the minimum data to decide with rigour. This is not bureaucracy — it is the argument you will need to present to the CEO and CFO when the time comes to approve the investment.
Do you know your current picking error rate, measured in errors per 1,000 lines picked? Have you calculated the cost per line — operator time, rework, returns? Do you know how many active references you have in stock right now? Does your ERP record warehouse movements in real time or only at the end of the day? Have you mapped the layout with fixed or dynamic locations? Do you know your weekly dispatch peak in lines per hour and the installed capacity of the team? Do you have a returns history broken down by reason — wrong reference, wrong quantity, packaging?
If you cannot answer three or more of these questions, the problem is not the software. It is the absence of baseline data. Solve that first — and do not underestimate the effort: in warehouses that have relied on manual picking for years, reconstructing this history can take four to six weeks of fieldwork.
Manual picking: what works and where it collapses
Manual picking works. In a distribution company with 800 references, two experienced operators and dispatch until 5 p.m., it works very well. The problem appears when volume, variety or deadline pressure grow — and in the distribution corridors from Lousada to Paços de Ferreira, that growth happens without prior warning and without a proportional scaling of the team.
The pattern we see repeated in INFOS projects is always the same. The company has a warehouse manager who knows the space by heart. He knows where each reference is, manages priorities off the top of his head, resolves dispatch conflicts over the phone. When that employee is absent, the warehouse slows down by 40%. When he leaves the company, he takes the warehouse with him — and the replacement takes six months to rebuild what was in his predecessor's head.
A warehouse that only works because João knows where everything is is not a managed warehouse — it is a hostage warehouse.
Manual picking also creates a silent regulatory problem. Without a digital record of movements, the batch-by-batch traceability required in sectors such as food, textiles with certified origin or footwear with international buyer compliance requirements becomes impossible to audit. It is not a matter of good practice — it is a matter of contract with the customer. When a German buyer asks for the batch history of a shipment from three months ago, the answer "we have it in the warehouse paperwork" is not good enough.
What changes with an integrated WMS
Integrated warehouse management is not installing a barcode scanner. It is linking the physical movement to the management system in real time — stock updated at the moment of scanning, not at the end of the shift. The concrete operational difference is this: the operator receives the picking list optimised by route, not by the order in which the order came in; each scan validates reference, batch and quantity before the product leaves the location; stock available for sale is updated in real time, which means the sales team stops promising what does not exist; returns are recorded with a reason and allocated to the correct batch; and the physical inventory ceases to be a traumatic annual event and becomes a rolling count by zone.
For those working in distribution, the integration between KORA Inventory Suite and ERP MULTI closes this cycle without manual intervention: the movement in the warehouse updates the stock in the ERP at the moment of scanning, and the customer order reflects real availability at the moment of confirmation — not last night's availability.
Decision matrix: manual picking vs. integrated management
| Criterion | Manual picking | Integrated management (WMS) |
|---|---|---|
| Lines/day volume | Up to ~300 lines with a stable team | Scalable without quality degradation |
| Number of active references | Works up to ~1,500 SKUs | Suitable from 500 SKUs with high turnover |
| Typical error rate | 1–3% of lines (varies with fatigue) | Below 0.3% with scan validation |
| Batch traceability | Impossible in real time; reconstructed after the fact | Automatic per movement |
| Dependence on a key person | High — undocumented tacit knowledge | Low — process codified in the system |
| Implementation cost | Nil (already exists) | Investment in software, hardware and training |
| Payback time | — | Typically 12–24 months in mid-sized distribution |
| Compliance with buyer requirements | Difficult to demonstrate through documentation | Auditable and exportable (SAF-T, batch reports) |
| ERP integration | Manual or end-of-day | Real time via API or native integration |
How to assess your situation in 4 steps
Measure the real cost of current errors. Take the returns from the last 6 months. Separate those caused by picking errors — wrong reference, wrong quantity. Multiply by the cost of collection, re-dispatch and customer credit. This figure is the business case. Without it, the conversation with the CFO stays in the realm of opinion.
Map the physical locations. Before any software, the warehouse needs fixed addresses — aisle, shelf, position. Without this, a WMS has nowhere to anchor movements. This step seems obvious and is systematically underestimated: in warehouses with 10 years of manual picking, there are always zones without an address, references stored "wherever there was space" and locations that exist on paper but not in practice.
Identify the 20% of references accounting for 80% of the volume. Optimising picking routes only makes sense if the highest-turnover references are in the most accessible positions. Do this analysis before defining layouts — and do not trust the warehouse manager's intuition without validating it against system data. The perception of what "moves most" diverges from the real numbers more often than one expects.
Validate integration with the existing ERP. An isolated WMS creates a second island of data — and two islands of data are worse than one, because they create divergences that nobody knows how to resolve. Integration with the central management system — whether via native API or via middleware — is a condition of success, not an option to consider in phase two.
Mistakes that cost projects
Implementing the WMS without revising the layout. The software optimises routes based on the current layout. If the layout is bad — and in most warehouses that have used manual picking for years, it is — the WMS optimises a bad layout with great efficiency. First revise the positions of the high-turnover references. Two weeks of physical reorganisation before go-live are worth more than six months of adjustments afterwards.
Digitising picking without involving the warehouse team from the outset. The warehouse manager who resisted the project for six months will hinder adoption — not out of bad faith, but because the old system was his power and his value. The way to win him over: involve him in designing the picking rules from the very first working session. When the rules are his, resistance turns into advocacy.
Buying hardware before validating the software. Industrial terminals, label printers and scanners have specifications that depend on the software — communication protocols, operating system requirements, compatibility with low-temperature or high-humidity environments. The correct order is always: software first, hardware specification next, purchase last.
Ignoring the initial inventory. A WMS starts with the stock that is in the system. If the stock in the system diverges from the physical stock — and it does, always — the system will manage the wrong stock with great efficiency. Carry out a full physical inventory before go-live. It is non-negotiable.
Treating the WMS as an IT project. Integrated warehouse management is an operational project with an IT component, not the other way around. The project lead must be someone with authority over the warehouse and over the dispatch processes. When the lead is the systems technician, the project delivers a system that works technically and that nobody uses correctly.
What the manuals do not mention
In distribution projects in Portugal, the greatest adoption risk is not technical. It is the interval between go-live and the moment the warehouse team trusts the system's stock more than the stock they see with their own eyes. That interval can last weeks or months, and during it operators do double-checking — system and manual count — duplicating the work and cancelling out part of the expected efficiency gain.
The way to shorten this interval is not additional training or internal communication. It is defining a divergence protocol from the first day of real operation. When the system says 12 units and the operator counts 10, what exactly is the process? Who records the divergence? Who authorises the stock adjustment? Within what timeframe? Without this protocol documented and communicated before go-live, the team resolves it on its own — and the system accumulates silent errors that only surface at the annual inventory, when it is too late to work out the origin.
There is a second detail that rarely appears in project documentation: the first week after go-live is the moment of greatest risk of informal abandonment of the system. If during that week there is a serious divergence that is not resolved quickly, the operators go back to paper — and regaining their trust afterwards is three times harder than having built it from the start. Intensive support in the first two weeks is not an implementation extra — it is the difference between a project that sticks and a project that stays on paper.
To explore the integration between warehouse management and the industrial value chain in more depth, the article KORA and ERP MULTI: control of the entire textile value chain shows how this link works in a sector with high reference complexity. If the question is broader — assessing whether the current ERP supports this growth — the article Industrial ERP Portugal: how to assess vertical fit before signing a contract offers a structured assessment method. For those still building the internal case for data centralisation, The importance of a centralising ERP for SMEs is the right starting point.
With margins so narrow in wholesale trade, the warehouse is not a secondary cost centre. It is where profitability is decided, picking line by picking line.
Sources
- INE — Instituto Nacional de Estatística. Estatísticas do Comércio 2024. Available at: www.ine.pt. (Data: overall trade margin in wholesale trade, 2024.)
Frequently asked questions
What picking error rate justifies investing in a WMS?
Rates above 1% of picked lines begin to compromise margin. With narrow trade margins in Portugal, each error generates a cost in rework, returns and customer credit. A WMS reduces this to 0.3% or less. If you have 500+ lines/day with an error rate above 1%, the investment typically pays for itself in 12–24 months.
Does manual picking work if we have few references?
It works up to around 1,500 SKUs with a stable team and controlled volume. Below 300 lines/day, the cost of implementing a WMS may not be justified. But if you have projected growth, high turnover or batch traceability requirements, start preparing now — migrating later is more expensive.
How long does it take to implement a WMS?
It depends on complexity. A basic WMS with barcode scanning integrated with the ERP takes 8–12 weeks in mid-sized warehouses. Before that, you need 4–6 weeks to validate baseline data — layout, active references, error history. Do not underestimate this diagnostic phase.
Does a WMS solve the problem of dependence on a key person?
Yes. The warehouse manager's tacit knowledge ("João knows where everything is") is codified in the system — locations, routes, priorities. When that person leaves, the process remains documented. The risk of a 40% degradation when someone is absent disappears.
What is the typical cost of a WMS for a mid-sized distribution company?
It ranges between 15,000 and 50,000 euros, depending on software, hardware (scanners, terminals), ERP integration and training. Add 10–15% for customisations. The return comes from reduced errors, rework and returns — not from staff savings, which are rare in distribution.
Can I implement a WMS without integration with the ERP?
Technically yes, but you lose the main benefit: real-time stock. Without integration, the WMS runs in isolation and needs manual or end-of-day synchronisation with the ERP. This keeps the risk of divergences. Native or API integration is essential so that the sales team sells what actually exists.
Is a WMS mandatory to meet traceability requirements?
For sectors such as food, textiles with certified origin or footwear with international compliance, yes. Without a digital record of batch-by-batch movements, you cannot audit or export histories. A German buyer will demand SAF-T or a batch report — warehouse paperwork is not good enough contractually.
What is the first step if I do not have baseline warehouse data?
Validate three essential metrics: current picking error rate (errors per 1,000 lines), cost per line (time + rework + returns) and returns history by reason. If you cannot answer these, the problem is not the software — it is the absence of data. Devote 4–6 weeks to reconstructing this before deciding on a WMS.