Five years ago, a garment manufacturer in the Vale do Ave region with 120 employees was processing 800 documents a day manually. They were losing 340 person-hours per month on classification, filing, retrieval and re-sending lost documents. When we proposed full automation, the response was immediate: "We can't stop." They were right. And we discovered it wasn't necessary.

The difference between a project that works and one that stalls is not the technology. It's starting in the right place, without stopping what already works.

The classic mistake: all or nothing

Most companies that embark on a document management project start with the right ambition but the wrong planning. They want to digitise everything at once: invoices, delivery notes, attendance sheets, approvals, contracts, technical reports. They set a cut-off date — "from 1 June, all paper is banned".

It works in 10% of cases. In the other 90%, the user goes back to paper because the system doesn't respond in time, because the integration with the ERP failed, or because nobody explained to them why the new screen is better than the file they've known for 12 years.

The project stalls. It costs money. It breeds resentment. Two years later, there are paper documents, digitised documents, documents in two places at once. Chaos.

We see this happen because document automation is presented as a technical problem when it is, in fact, an operational problem.

Start where paper hurts most

The strategy that works is the reverse: identify the process that causes the most friction, the most wasted time, the most errors. Not the most critical from a strategic point of view — the one that hurts most day to day.

In a garment manufacturer, it's often procurement. The delivery note arrives, the warehouse manager marks the receipt in the ERP, but the PDF file gets lost in an email folder. When the supplier questions the quality of a batch in September, no one can retrieve the original signed note. Result: an argument over the receipt date, a delayed return, stock sitting idle.

In distribution, it's picking. The operative prints the list, ticks off manually what they've collected, and the list ends up on the warehouse floor. When there's a discrepancy between what was sent and what was received, there's no record of who collected it, when, or in what quantity.

In a metal/plastics factory, it's the overtime approval workflow. The sheet passes through 4 people (supervisor, production manager, HR, accounting), takes 6 days, and gets stuck on someone's desk. By the time it arrives the following month, the context is gone.

Document automation is not an IT project. It's the operational answer to a problem that already exists and costs money every single day.

Choose that process. Not the smallest — too small doesn't justify the investment. Not the most complex — too complex will require customisation and will take time. Choose the one that directly affects operational efficiency and has a measurable financial impact.

Implement only in that flow. Capture the source document (invoice, note, timesheet). Classify it automatically with cognitive capture (intelligent OCR). Route it to the right person. Create an audit trail of who saw it, when, and how long it took. Integrate with the ERP if necessary.

This takes 8 to 12 weeks. The rest of the operation feels nothing.

Silent integration with what already exists

The second mistake is assuming that document automation requires a complete new system. Often, it doesn't. What works is integration with the ERP they already have — MULTI, SAP, Primavera — and with the processes that already work.

A factory that has had MULTI for 10 years doesn't need to replace MULTI. It needs a module that captures the delivery note in PDF, extracts the serial, quantity and batch numbers, validates them against the order in the ERP, and automatically sends them to the warehouse. This runs in parallel with MULTI. MULTI doesn't know there's automation. Neither does the user — they see the document in the system, and that's it.

This is possible because modern document automation is modular and ERP-agnostic. It doesn't replace. It reinforces.

Many IT directors are afraid of integration because the last integration they did was a disaster. Two databases, three vendors, 6 months of delays. But this is different. Document automation doesn't touch the core of the ERP. It feeds it, but doesn't replace it. And if something goes wrong, paper still works as a fallback — for 2-3 weeks, while it's fixed.

What really takes time (and it's not the technology)

If you see a document automation project with a 14-week duration, the first 4 weeks are technology. The other 10 are people.

Process mapping: 2 weeks. Meetings with the people who do the work (not those who supervise it) to understand the edge cases — why some invoices take 3 days to reach their destination, why some documents pass through 6 people and others through 2.

Testing with real users: 3 weeks. The warehouse manager with 18 years at the company will find 12 scenarios no analyst anticipated. This is expected. This is good.

Training and onboarding: 2-3 weeks. It's not a 40-page manual. It's the supervisor showing the operative how the document now arrives on their screen, and they click to confirm, and it's done. Finished.

This is the reality. The technology is the easy part.

Metrics that matter (and they're not the ones you think)

When the project ends, there's always the temptation to measure "percentage of documents digitised" or "paper reduction". This is irrelevant. Paper doesn't disappear. Some documents will remain on paper for legal reasons, or because the user prefers it, or because the customer sends them on paper and there's no reason to change.

What matters is: how long does a document now take to go from A to B? How many people have to touch it? How many times is it re-typed? How many errors of location or loss?

A footwear factory that automated its overtime approval workflow went from 6 days to 1 day. This isn't spectacular — it's operational. But operational is all that matters. The payroll sheet arrives on time, without errors, without argument. HR can respond within 24 hours to questions about hours worked. Accounting receives everything validated. This costs implementation time, but saves time permanently.

Measure this. Not paper.

Document automation that doesn't reduce cycle time or operational error is merely computerisation. Computerisation isn't worth the cost.

Where to truly start

Three practical steps, not generic ones:

One: Meet with the people who do the work. Not managers. The warehouse manager, the accounting assistant, the production supervisor. Ask them: which document bothers you most to process? Which one makes you lose the most time? Which one forces you to make phone calls because you don't know where it is?

Two: Calculate the impact. How many documents of that type per month? How long does each one take? If it were halved, how much would you save in person-hours? That's your ROI. It's not software — it's operational time freed up.

Three: Prototype with a vendor that understands your industry. Don't buy licences. Ask for a 4-week proof of concept (PoC). Let the user touch it, make mistakes, suggest changes. If they say "this works", it's because it works. If they say "this is complicated", go back to paper — it costs 4 weeks, not 14 months.

This is starting without stopping the operation. This is document automation that makes sense.

Regulatory framework: what compels action

If you don't yet have document automation, there are three regulations that will force the decision.

Ordinance 195/2020 requires the monthly reporting of SAF-T to the Tax Authority. This means that all sales, purchase and inventory documents must be mapped, auditable and synchronised. If you have invoices on paper, notes in email, and misaligned records in the ERP, the Tax Authority will find discrepancies. This is a risk.

The NIS2 Directive (transposed into Portuguese law via DL 65/2025) requires medium-sized companies to implement cybersecurity controls. This includes traceability of document access, encryption, backup, and auditing of who touched what. Paper doesn't offer this. Document automation with ISO 27001 does.

eIDAS 2 (qualified digital signature) allows digitally signed documents to have legal value equivalent to paper. This changes the game. A digital approval with a timestamp is more secure and more auditable than a paper signature.

These regulations are not optional. And implementing them without document automation is impossible.

Integration with the ERP: the practical path

If you have MULTI, the integration is direct. MULTI has document management and workflow modules that feed off automatic capture. If you have another ERP, document automation works as a pre-processing layer — it receives the document, validates it, classifies it, and passes it to the ERP with data already structured.

This reduces manual entry errors. A delivery note that arrives in PDF is captured automatically, validated against the order, and entered into the ERP without anyone having to type. This is a direct gain in time and quality.

For companies with multiple branches or partners, MULTI Connect synchronises documents across locations. A branch that receives a note can validate it and send it to the branch of origin without email, without paper, without delay.

Where not to start

Avoid implementing document automation in processes that aren't yet stable. If your overtime approval workflow changes every month because HR doesn't know what it wants, automating will freeze the chaos. First stabilise the process. Then automate.

Also avoid choosing a document that is very specific to your company. If it's an internal form that only you use, the customisation will be expensive. Choose documents that are standard in your industry — invoices, notes, timesheets, purchase orders.

And above all, avoid choosing a vendor with no experience in Portuguese industry. Document automation that works in Lisbon consultancies doesn't work in a textile factory in the Vale do Ave. The context is different. The documents are different. The users are different.

Next steps

If you've reached this point and recognised your problem — documents that take too long, that get lost, that cause errors — contact an INFOS specialist. Not to buy software. To understand what your real pain point is and how to start without stopping the operation.

A 4-week proof of concept costs less than a month of wasted time on poorly processed documents. And if it doesn't work, you go back to paper. Simple.