Five years ago, a textile company with 120 employees in the Vale do Ave asked us to "go paperless". It was spending 14 hours a week on invoice approvals, warehouse dispatch reports and paper timesheets. The CEO wanted to comply with DL 28/2019 and save money at the same time. We implemented a document management system. The result: we brought operations to a standstill for 11 days. The warehouses did not know how to confirm picking. The production supervisors could not sign work orders. IT had to roll back.

Today we see things differently. Paperless is not a technology project. It is an operations project with technology. And that difference changes everything.

DL 28/2019 does not force you to get rid of paper

First uncomfortable truth: Decree-Law 28/2019 (electronic invoicing, ATCUD, SAF-T) does not prohibit paper. It only prohibits not having digital traceability of operations. You can have a paper document — provided you have a certified, auditable digital copy integrated into your ERP.

This changes the whole conversation. It means you do not need to eliminate paper overnight. You need to create a parallel digital flow that coexists with paper until paper is no longer needed. And that coexistence is where most companies fail.

The mistake is not keeping paper. The mistake is having paper and digital in parallel with no connection between them.

A garment factory in Famalicão with 180 workers has daily production sheets. Each sheet is signed, stamped, photographed and stored in a box. In an AT audit, that paper is the proof of everything. But when the ERP needs the data, it is keyed in again manually. Two truths, zero connection. This is what brings things to a standstill.

DL 28/2019 requires you to have AT certification in your invoicing software. But it does not require your warehouse, production or HR to be digital. It requires you to be auditable. A critical difference.

Why "fast" implementations fail within 30 days

We see projects that promise "paperless in 60 days". They buy document management software, run two training sessions, switch off paper, and wait. Reality arrives on the first Monday of normal operation.

A warehouse works to a rhythm. Picking, checking, packing, dispatch — all in quick sequence. If the document capture system (scanner, mobile terminal, digital signature) is not integrated into that rhythm, the worker goes back to paper. Not because they are stubborn. Because paper is faster.

The successful implementations we have seen share a pattern:

  • Week 1-2: an exact diagram of the document flow (where it originates, who it passes through, where it ends).
  • Week 3-4: a pilot covering 20% of operations, with paper still active in parallel.
  • Week 5-8: time measurement (before: 4 minutes per document; after: 3.2 minutes). If the time gets worse, the digital process is not transparent. Go back.
  • Week 9-12: phased expansion with ongoing training.
  • Week 13+: elimination of paper only when digital is faster.

This takes 3 months. Not 60 days. The difference is that at 3 months it stuck.

What kills fast projects is the illusion that technology changes behaviour. It does not. Behaviour only changes when the new way is visibly better. If the digital signature is more clicks than paper + pen, people go back to paper. If the scanner is far from the workstation, they go back to paper. If the system goes down and paper is the plan B, paper stays.

Integration is the real name of the game

A construction materials distributor in Lousada has 45,000 SKUs. When a picking is confirmed on paper, someone has to key it into the ERP. When a return is signed on paper, someone has to create the credit note. When a payslip is printed, someone has to file it in a folder.

This is invisible work that no one accounts for until they try to go digital.

When you implement document management well, what changes is not that paper disappears. It is that the document's information goes straight into the ERP. Picking confirmed on a tablet? It is already in the ERP. Payslip signed digitally? It is already in the legal archive and linked to the employee. Invoice received from a supplier? Captured by OCR, validated, and ready for automatic approval if it meets the criteria.

This requires your ERP and your document system to talk to each other. Not "integration" in the consulting sense (which is always expensive and slow). Technical integration: APIs, webhooks, automated workflows.

A document with no integration into the ERP is just a nice archive that nobody uses.

We see companies with DocuShare or Tungsten (solid platforms) but with no connection to the ERP. The document sits in the archive. Operations stay on paper. The best of both worlds becomes the worst of both worlds.

The real test: how much time do you actually gain?

Before you start, work this out: how many hours a week do you spend on manual operations that a smart digital document could eliminate?

  • Supplier invoice approval (manual checking + approval routing)?
  • Signing work orders or production sheets?
  • Picking confirmed on paper that is then keyed in?
  • Managing payslips or expense receipts?

If the answer is less than 5 hours a week, paperless is a luxury. If it is more than 15 hours a week, paperless is business. If it is 20+ hours a week, it is urgent — and it means your operation is losing money you cannot even see.

A well-executed implementation (3 months, internal team + external consulting, 15,000-20,000 euros) pays for itself in 6 months if you gain 10+ hours a week. In 12 months if you gain 5 hours a week.

But this only works if the integration is real. If it is just a nice digital archive, you never get it back.

The Portuguese regulatory context (and why it matters)

DL 28/2019 and Ordinance 195/2020 (monthly SAF-T communication) created a clear obligation: you must have digital traceability of financial operations. The AT audits this. Full stop.

But there are two levels of compliance: the minimum (paper + photograph + folder) and the smart approach (natively digital + integrated + automatically auditable).

If the AT audits you today, the minimum passes. If it audits you in 2 years and the law tightens (as NIS2 is doing with cybersecurity), the minimum fails. Because NIS2 requires digital traceability of access, modification and removal — paper leaves no trail.

This means "going digital" is not optional in 2025. It is strategic.

Sectors such as textiles (ATP) and footwear (APICCAPS) are starting to require traceability of production documents from suppliers, driven by pressure from international brands (Inditex, Decathlon, Lacoste). Paper no longer passes. Certified digital documents do.

Questions you must ask before you start

Before contacting a software vendor, do this internally:

  • Which document takes the longest to process? How many people handle it? How long does it sit waiting?
  • Does your ERP have an API for document integration? (Check with your vendor — many say they do, but it is limited.)
  • Do you have the staff to manage the transition? Or do you need external consulting? (Consulting costs money, but it avoids a standstill.)
  • What is your plan B if the digital system goes down? (Does paper still exist? A clear procedure?)

If you cannot answer these, you are not ready for paperless. Wait another 3 months, set up an internal working group, and get ready.

What we see working

Companies that gain time with paperless have this in common.

First, they start with one process, not the whole company. A plastic injection factory in Marinha Grande started with work orders. It moved from signed paper to a tablet with digital signature + photo of the part. It gained 6 minutes per order (from 12 to 6). It expanded to picking. Then to production sheets. It took 4 months to get everything integrated. Today, 8 hours a week gained.

Second, they keep paper as plan B for 3 months. This is not weakness. It is engineering. If the system goes down, paper ensures operations do not stop.

Third, they measure time before and after. Not "feelings". A stopwatch. If the time does not improve within 4 weeks, they review the flow. If the time gets worse, they roll back.

Fourth, they train obsessively. Not a 2-hour session. Ongoing training. Every new worker who joins goes through 30 minutes of hands-on support with the document system. Every 2 weeks, a 15-minute meeting with the section supervisor to resolve friction.

Document capture: paper, scanner or OCR?

There are three ways to get a document into the system:

Photographed paper. The worker takes a photo with a mobile phone and sends it to a folder. Then someone digitises it. Slow, error-prone, but it works as a plan B.

Dedicated scanner. A mobile terminal or fixed scanner in the warehouse. The worker scans the document and it goes straight into the system. Fast, but it requires hardware and discipline about placement.

OCR + cognitive capture. The document comes in (paper or PDF), the system extracts the data automatically (names, values, dates, signatures), validates and routes it. More expensive, but it eliminates 80% of the manual work.

Most successful implementations start with a scanner (weeks 1-4), then move to OCR when the volume justifies it (weeks 5-12).

Digital signature: eIDAS, qualified certificate, or timestamp?

If the document has to be legally valid (invoices, work orders, production sheets), you have to use a qualified digital signature under the eIDAS Regulation. This means a digital certificate issued by an accredited certification authority (such as the Citizen Card or platforms like DocuSign with PT certification).

Paper + pen is not legally valid in an audit if the law tightens. A qualified digital signature is. A difference of 2 years of the chain if you falsify it.

Many systems use a timestamp instead of a qualified signature. It works for traceability, but not for legal validity. Check with your legal adviser what the exact requirement is for your sector.

Integration with the ERP: API, webhooks, or manual synchronisation?

Once the document has been captured, it has to reach the ERP. There are three routes:

Real-time API. The document system sends data straight to the ERP (picking confirmed → immediately in the ERP). Fastest, zero manual errors, but it requires the ERP to have an open API (many generalist ERPs do not).

Webhooks with a queue. The document sits in a queue and the document system notifies the ERP periodically (every 5 minutes, every hour). A middle ground between speed and simplicity.

Manual or batch synchronisation. Someone exports data from the document system and imports it into the ERP. Slow, error-prone, but it works if the volume is low (fewer than 50 documents a day).

If you choose an API, check with your ERP vendor whether the API is documented, whether it is supported, and whether there is an additional cost. Many sell "integration" as if it were free, but charge for it later.

Once processed, the document has to be stored legally. There are three options:

Native ERP. Some ERPs have an integrated document management module. Advantage: everything in one place. Disadvantage: weak archive, no OCR, no advanced workflow.

DocuShare (Xerox). A dedicated archiving platform. Strong on compliance, legal retention and search. Expensive (10-20K€/year). Good for companies with many documents (100K+/year).

Tungsten (ABBYY). A cognitive capture platform. Strong on OCR, data extraction and automation. Mid-range cost (5-15K€/year). Good for companies that want to eliminate manual work at the input stage.

Most Portuguese SMEs start with Tungsten (capture + archive) and then integrate with the ERP.

Approval workflow: automatic or manual?

Once captured and validated, a document (invoice, work order, timesheet) needs approval. There are two models:

Manual approval. The document sits in a queue, the supervisor opens it, reads it, approves or rejects it. Time: 2-4 hours (if the supervisor is present). Risk: the queue grows, documents are delayed.

Automatic approval with rules. If it is an invoice from a regular supplier, with a value within budget and the correct date → it is approved automatically. If it does not meet the criteria → it goes to a manual queue. Time: 5 minutes (automatic) + 1 hour (manual if needed). Risk: zero, because the system validates beforehand.

Automatic approval with rules is where we gain the most time. A factory that received 200 invoices a month went from 40 hours a month of manual approval to 8 hours a month (only those that do not meet the criteria).

Training: how to make sure nobody goes back to paper

Training is where most implementations fail. It is not for lack of willingness. It is for lack of time and repetition.

A model that works:

Week 1: group training. All users (warehouse, production, administrative) watch a 1-hour demo. They see the old flow (paper) vs the new one (digital). They see how much time they gain. This creates buy-in.

Week 2-3: hands-on training. Each user spends 30 minutes with the system (capture, signature, confirmation). With real data (the company's documents). With the section supervisor alongside them.

Week 4-8: intensive support. Every day, someone from the implementation team is on the shop floor. They resolve problems in real time. They document friction. They adjust the flow.

Week 9+: remote support. Helpdesk by phone or email. Response within 2 hours.

Each month: review meeting. Section supervisors + IT + implementer. Metrics (time gained, errors, rejection rate). Adjustments.

Plan B: what to do if the system goes down

No system is 100% reliable. Servers go down, the internet goes down, software has bugs. If your document system goes down and operations cannot function without it, you have a problem.

A plan B that works:

  • First 3 months: parallel paper. If digital goes down, you go back to paper. The document is then keyed in when the system comes back.
  • Month 4+: paper only as a last resort. But with a clear procedure: which form, who signs, where it is stored.
  • Always: data backup. Daily. Tested monthly (it is not a backup if you do not test the restore).
  • Always: redundancy. If the server goes down, there is another that takes its place (automatic failover).

A distributor in Paços de Ferreira had a 4-hour outage. Because it had a plan B (paper), operations did not stop. When the system came back, they keyed in the documents from paper. Loss: zero.

Metrics: how to measure success

Before you start, define metrics. Not "feelings". Numbers.

  • Processing time. Before: 4 minutes per document. After: 2.5 minutes. Target: 2 minutes. Measurement: stopwatch, 10 documents a week.
  • Error rate. Before: 3% of documents with errors (manual keying). After: 0.2% (OCR + validation). Target: 0%. Measurement: monthly audit of 100 documents.
  • Rejection rate. Before: 5% of documents rejected for missing signature/date. After: 0.5% (the system requires it before confirmation). Target: 0%. Measurement: system log.
  • Approval time. Before: 24 hours (queue). After: 2 hours (automatic + manual). Target: 1 hour. Measurement: entry/exit timestamp in the system.
  • Hours gained per week. Before: 14 hours (keying, filing, searching). After: 6 hours. Target: 4 hours. Measurement: monthly user survey + observation.

If no metric has improved by 2 months, something is wrong. It may be the flow, it may be the training, it may be the integration. Investigate before continuing.

The real cost of an implementation

There is no such thing as a "cheap implementation". There are implementations that pay back the investment and implementations that do not.

Typical cost (SME with 50-200 employees):

  • Software (document management + integration): 8-15K€ (year 1).
  • Hardware (scanners, mobile terminals): 3-8K€ (one-time).
  • Consulting (3 months, 1 consultant): 12-18K€.
  • Internal training: 2-4K€ (staff time).
  • Total: 25-45K€ (year 1).
  • Year 2+: 8-15K€ (licences + support).

If you gain 10 hours a week (cost: 15€/hour = 150€/week = 7.8K€/year), you pay it back in 3-4 years. If you gain 20 hours a week, you pay it back in 1.5-2 years.

This is an investment. Not an expense.

When NOT to go paperless

Paperless is not for everyone. If you are in one of these situations, wait:

Sources

  • Decree-Law no. 28/2019, of 15 February — Legal regime for electronic invoicing, ATCUD and digital traceability obligations (Diário da República, Series I)
  • Ordinance no. 195/2020, of 28 August — Monthly communication of the SAF-T file (Diário da República, Series I)
  • Tax and Customs Authority (AT) — Practical guide to electronic invoicing and software certification requirements (portal.at.gov.pt)
  • ISO/IEC 27001:2022 standard — Information security management in document systems and digital archives
  • CNPD (National Data Protection Commission) — Guidelines on the processing of personal data in document management systems (cnpd.pt)