An industrial SME does not grind to a halt for lack of orders. It grinds to a halt because growth exposes what Excel was hiding: stock that nobody can locate, production costs calculated on instinct, a month-end close that drags on for twelve days. The most expensive misreading we see is exactly this — management concludes it needs "new software", when what it has is a data problem that no software solves on its own.
So, the argument, without beating about the bush: the problem is almost never the old software. It is that scaling without a single core of data multiplies the error, not the output. A factory that doubles its volume with manual processes doubles its points of failure too. The ERP is not there to "modernise" — it is there so that the twentieth new customer does not break the operation that worked with ten. This piece gives you a grid to work out whether you are ready, and what to audit before you sign.
And there is plenty of room to grow in that direction. In 2025, only 53.7% of companies in Portugal with 10 or more people used enterprise management software, according to INE. Close to half of the business fabric still runs its operation on spreadsheets and well-informed heads — which works beautifully until the well-informed head takes its holidays in August.
What you need before you start
Before looking at any demo, have this in place. Without these prerequisites, the project starts off crooked — and crooked cannot be straightened halfway through the migration, it can only be straightened by starting over.
- A map of the real processes — not the ones in the manual, the ones that happen on the Gemba.
- The owner of each process identified by name, not by department.
- A clean item master file (or an honest awareness of how dirty it is).
- Monthly transaction volume: orders, delivery notes, invoices, production orders.
- Confirmed legal requirements: SAF-T, ATCUD, AT certification, eIDAS.
- A budget with an annual support allocation, not just a start-up one.
- A sponsor in management who can withstand twelve months of friction.
Diagnose where growth hurts
The symptom is not the problem. "The close takes too long" is a symptom; the cause is usually that purchasing, production and finance do not share the same item cost. Walk through the operation and list where data is re-entered by hand.
In projects we have supported, the pattern repeats itself with almost comic stubbornness: the same item serial number has three different formats in three applications. In the sales system it is REF-0455, in the warehouse it is 455, on the foreman's cost sheet it is ref455 azul. No one is wrong — each solved their own problem on their own screen. The ERP does not reconcile this by magic. It forces you to decide which is the true format, and that decision, taken in a room with the warehouse manager and the purchasing manager arguing it out, is worth more than any licence.
Three questions cut to the bone. Where is the same piece of data typed in more than once? Which management decision waits days for a report done by hand on a Saturday? And which process depends on one person who, if they leave, takes the knowledge out the door with them? If you answered those three with concrete names, you already know where the ERP will hurt first — and where it will be worth most.
Demand vertical fit, not generic flexibility
A general-purpose ERP promises it can be configured for everything. In practice, it forces you to mould the operation to the software. In a footwear collection with 800 to 1,200 SKUs and three axes — colour, size and last — an ERP that does not model the colour-size matrix natively turns every order into a headache. And it is not an abstract headache: it is the parent company's buyer visiting in February for the women's collection, asking for the same reference in four widths, and the system demanding that a new item be created for each combination. Multiply by a thousand.
The same in the textiles of the Vale do Ave. When the parent company asks for compliance with the EU Strategy for Sustainable and Circular Textiles, batch-to-batch traceability stops being a luxury. The dyer asks which supplier the yarn in this batch came from, and the answer has to come out of the system in seconds, not from a notebook in the foreman's drawer. ERP MULTI builds five specialised verticals on a common core, precisely because the logic of an auto-parts warehouse is not that of a finishing house. Before deciding, read how to assess vertical fit before signing a contract.
In the demo, do not accept the vendor's default scenario. Bring your most complicated item — the one with a special last, a made-to-measure colour and a unit of measure in pairs and in pairs-per-box at the same time. If the salesperson needs to "go back to the office to configure it", you have your answer.
- Does the ERP model your real unit of measure (pair, metre, roll, kg) without contortions?
- Does it support the colour-size-last matrix in the core, or does it hang an external plugin on it?
- Is batch-to-batch traceability built in from the ground up, or is it a bolted-on module that nobody can guarantee in an audit?
Start modular, not all at once
The mistake that ruins budgets: wanting to connect production, HR, finance, stock and sales in the same go-live. The modular architecture exists for phasing. Activate the core that stems the biggest haemorrhage first — usually stock and invoicing — and only then expand.
A textile factory of around 80 employees does well to stabilise finance and stock over six months before touching the shop floor. Connecting real-time production capture and OEE to a core that is not yet the source of truth produces just one thing: pretty dashboards with wrong data, presented with every confidence at the management meeting. A wrong number on a big screen is more convincing than a right number on a spreadsheet — and that is where decisions come off the rails.
Treat master data as a project in its own right
No ERP makes up for an item file with 40% duplicates. Data migration is where most delays are born — not in the technology. Appoint someone dedicated solely to cleaning items, customers, suppliers and technical lists before any load. This person is not an intern with a free weekend; they are someone who knows the business well enough to know whether Malha 30/1 and malha 30-1 penteada are the same item, or not.
The ERP does not clean your data. It forces you to admit how dirty it was — and that admission, made early, is what separates a 14-week go-live from a 14-month nightmare.
Decision matrix: are you ready for an ERP?
Score each criterion from 1 (does not exist) to 3 (mastered). Below 18 points, sort out the prerequisites before proceeding. Do this in a one-hour meeting with the CFO and the IT manager — not on your own, and not in Excel three months away.
| Criterion | 1 — Fragile | 3 — Ready |
|---|---|---|
| Documented processes | Only in people's heads | Mapped and with an owner |
| Master data quality | Duplicates and various formats | Single, validated file |
| Sector vertical fit | Generic ERP moulding the operation | Vertical models the operation |
| Management sponsorship | Delegated to IT | CEO/CFO committed |
| Legal compliance | To be confirmed | SAF-T, ATCUD, AT guaranteed |
| Support allocation | Start-up budget only | Annual maintenance planned |
| Planned phasing | Everything big-bang | Modules by priority |
Common mistakes and how to avoid them
Some stumbles show up in almost every project we have set foot in, and none of them is technical. They are human, and therefore more stubborn.
Buying by the feature list. The "yes/yes/yes" table in the brochure proves nothing. Ask for a proof of concept with your real data and your most complex scenario — the order that usually causes trouble, not the one the salesperson picks.
Underestimating the warehouse manager. In the Lousada/Paços corridor, the person who decides the success of the onboarding is the one who does not put the radio down for two hours. If the rollout takes him off the ground on a dispatch morning, he sabotages it — not out of malice, but out of operational survival. Involve him in the design, not at the end, and show him the new picking screen before making him use it in front of the team.
Ignoring the IT hero. The self-taught type with 15 years of business knowledge and no degree is either your greatest ally or your greatest brake. He knows every exception the manual does not foresee. Give him technical prominence and he opens every door for you; go over his head and he remembers everything that is going to go wrong — and he is right.
Taking funding for granted. The PRR reserves 650 million euros for the digital transition of companies under the Empresas 4.0 component, according to the Government of Portugal. But the money gets stuck in poorly prepared technical reports. Treat the application with the same rigour as the project — the difference between approved and "returned for clarifications" is usually the quality of the technical description, not the merit of the investment.
Not measuring the before. Without a baseline of close days, stock-outs or OEE, you never prove the return. Record the current numbers this week, before memory prettifies them — a year from now nobody remembers that the close took twelve days.
Where this leaves you
Take the matrix, score the company in that one-hour meeting, and decide which module hurts most. Then cross that priority with the analysis of how to turn production data into decisions — because an ERP without reading in Qlik Sense still hides what growth is going to cost.
The software is the easy part. The hard part is admitting, in a room with the CFO and the IT hero, that the data was dirtier than anyone wanted to say out loud. Those who make that admission early get going in fourteen weeks. Those who put it off pay for it in fourteen months — and, to boot, with pretty dashboards lying along the way.
Sources
- INE — Survey on the Use of Information and Communication Technologies in Enterprises, 2025.
- European Commission — Digital Decade Report, 2025.
- Government of Portugal — Recovery and Resilience Plan (PRR), Empresas 4.0 component, 2021.
- Eurostat — Labour productivity per hour, 2022.
- ATP / INE — Textile and clothing industry exports, provisional data, 2026.