Every month the same ritual plays out in a factory HR office: someone exports the clock-ins from the time clock, pastes them into an Excel file, cross-checks them against the shift schedules, corrects the absences by hand, and only then sends the file off for payroll. Two people, three days, one careless error that blows up on payday. This guide gives you a four-stage phased adoption plan to automate HR without breaking legal compliance — in the right order.
The thesis is simple and uncomfortable: most Portuguese companies do not have an HR software problem — they have a start-up sequence problem. They buy the whole suite, try to connect everything at once, and the project dies in the recruitment module while attendance still lives in Excel. HR automation does not fail for lack of features. It fails from an excess of ambition in the first quarter.
And the context is tightening. In 2024, 65% of employers in Portugal reported difficulty finding professionals with the profile they need — one of the highest figures in the world (ManpowerGroup, 2024). When talent is scarce, every error on the payslip and every wrongly recorded day off cost trust that is no longer easy to recover.
What you need on the table before you start
Before looking at any screen, take an honest inventory of what you have. Inventory all the sources of attendance data — time clocks, shift sheets pinned to the wall, paper holiday charts that the supervisor scribbles on with a ballpoint. Identify the current payroll system and, above all, the exact file format it consumes; this is where 80% of integrations get stuck. Write down the monthly declarative obligations — the Monthly Remuneration Declaration (DMR) to the Tax Authority and Social Security — and record the handling of employees' personal data in accordance with the GDPR and Law 58/2019.
Two things the manuals forget. First: map who actually decides — in Portuguese family businesses it is almost always the trio of CEO + CFO + head of HR/IT, and that IT person is often a self-taught individual with fifteen years of business knowledge in their head and no diploma on the wall. Win them over early or the project stalls. Second: if you have 50 or more employees, confirm that the whistleblowing channel under Law 93/2021 is already operational — it is not pure HR, but it falls into the same bucket of labour compliance that you will audit.
Step 1 — Core first: the single dossier before anything else
Before automating anything at all, there has to be a single source of truth about each employee. Contract, category, remuneration, training, documents. This is what pplPortal pplCore is: the single dossier from which everything else feeds.
The mistake we see in HR projects is starting with the shiny feature — AI recruitment, colourful dashboards — when the base data is still scattered across three spreadsheets with different versions of the same name. Do not move forward without a stable Core. Migrate each employee's data into a single record, with no duplicates. Define who accesses what — the GDPR requires access minimisation, and "everyone sees everything" does not pass an audit. And validate that the legal documents, from the contract to the copy of the identity document, are associated with the right record. It seems bureaucratic. It is the concrete foundation that stops the building settling crooked further down the line.
Step 2 — Evolution: attendance that feeds the salary
Here is the fastest return of all. pplEvolution handles attendance, schedules and shifts, and connects directly to payroll. The file is no longer assembled by hand.
In a typical garment workshop in the North, with rotating shifts and seasonal overtime when the parent company's collection comes in, manual overtime calculation is where half of the payday complaints are born. A worker who did 14 hours of overtime and sees 11 on the payslip does not open a polite ticket — she goes to the supervisor, and the supervisor goes to HR, and the morning is lost. Automate that calculation and you gain two things: time and industrial peace. For the before/after in detail, the article payroll processing: Excel vs. integrated software breaks down the hidden cost of the spreadsheet.
Three golden rules at this stage. Configure the shift and hours-bank rules once, not every month. Run a complete processing cycle in parallel with the old method before turning off Excel — a whole month, value by value. And check that the values for the DMR come out correctly, always with a window for human review before submitting.
Step 3 — Advanced: performance, skills and training
With the base solid and attendance automated, in comes the layer that retains people. pplAdvanced manages performance, the skills matrix and training plans. This is the part that the manuals treat as an "extra" and that, in practice, is what curbs turnover.
In 2023 the average voluntary turnover in Portugal was 10.6%, and 52% of companies admitted difficulty retaining talent (Mercer, 2023). In a footwear factory in Felgueiras, losing an experienced pattern maker in mid-August — when the international buyers arrive to see the menswear collection — is not an HR figure: it is an order at risk and a sample that no one knows how to finish. Mapping skills and planning succession stops being a luxury. It is knowing, before the pattern maker leaves, who on the team can hold the fort during the peak season.
Automate the Core and attendance first. Performance and recruitment are only worth anything once the base data is already trustworthy.
Step 4 — Talent + AI: recruitment and intelligent assistant
Only at the end do you connect recruitment (pplTalent) and the AI layer. And here a non-negotiable principle applies: AI in HR must be explainable, with no black box, aligned with the GDPR and with the risk classification of the AI Act (Regulation EU 2024/1689). A turnover-risk suggestion that no one can justify is not a tool — it is a legal exposure waiting to happen.
The generative AI component is accessible via natural language, including via WhatsApp, for the employee to check their holiday balance or request a payslip without opening a ticket. It is the difference between an HR that responds and an HR that chases paper — especially in an ageing workforce that will not install a new app to see how many holiday days they have left, but sends a message without a second thought. The guide pplPortal and AI in HR goes deeper into how to digitalise without losing people's trust.
| Module | Solves | When to connect |
|---|---|---|
| pplCore | Single dossier, fragmented data | Stage 1 — foundation |
| pplEvolution | Attendance, schedules, payroll connection | Stage 2 — quick return |
| pplAdvanced | Performance, skills, training | Stage 3 — retention |
| pplTalent + AI | Recruitment and intelligent assistant | Stage 4 — when the base is trustworthy |
The five mistakes that derail the project
- Connecting everything at once. Follow Core → Evolution → Advanced → Talent. Each stage validated before the next, with no exceptions for calendar pressure.
- Turning off Excel before running a parallel cycle. One month of duplicate processing, compared value by value. It is tedious. It is cheaper than correcting 40 payslips in September.
- Leaving the shift rules in one person's head. Document schedules and the hours bank in writing before the configuration — not "the rules Manuel knows by heart", because Manuel retires and takes them with him.
- Adopting AI without explainability. Require that every AI suggestion be justifiable and traceable, for GDPR and AI Act compliance.
- Ignoring the person clocking in on the shop floor. Adoption is measured there, not in the meeting room. Involve the operational team early, or you will have a perfect system that no one uses.
Where to start this very week
Write your four-stage phased adoption map and mark which module solves this quarter's most expensive pain — it is almost always attendance connected to salary, because it is the one that bleeds time every month and the one that ruins trust when it goes wrong. The order is not an aesthetic detail: it is the difference between a project that delivers value by the second stage and one that gets stuck in recruitment with the base still in spreadsheets. To go deeper, see how to retain talent through efficient HR processes and how to monitor all of this in a single dashboard in business KPIs for industrial directors.
Sources
- ManpowerGroup, Talent Shortage Survey, 2024 (talent shortage in Portugal).
- Mercer, turnover and retention data in Portugal, 2023.
- Tax Authority and Social Security — Monthly Remuneration Declaration (DMR).
- Regulation (EU) 2016/679 (GDPR) and Law No. 58/2019, of 8 August.
- Law No. 93/2021, of 20 December — general regime for the protection of whistleblowers.
- Regulation (EU) 2024/1689 (AI Act) — risk classification of AI systems.
Frequently asked questions
Is pplPortal legally mandatory in Portugal?
There is no legal obligation to use any specific software. However, it is mandatory to comply with the legal declarations (DMR, Social Security), record attendance and ensure GDPR compliance. pplPortal facilitates this compliance, but any system that meets these requirements is valid.
How long does implementing pplPortal take in a company with 100 employees?
The recommended phased implementation takes 4-6 months: Core (1-2 months), Evolution/attendance (1 month), Advanced (1-2 months), Talent (1 month). The time varies according to the quality of the existing data and the availability of the internal team for migration and testing.
Can I implement just attendance and leave Excel for the rest?
Technically yes, but it is not recommended. Attendance without a solid Core generates loose data that does not integrate with payroll or compliance. The phased sequence (Core → Evolution → Advanced → Talent) ensures that each module feeds the next without errors.
Does pplPortal comply with Law 93/2021 on whistleblowing channels?
pplPortal is not a whistleblowing channel in itself. However, it integrates with labour compliance systems. Companies with 50+ employees need an operational channel in accordance with Law 93/2021, which must be separate but audited within the same compliance context as HR.
What is the risk of migrating data from Excel to pplPortal?
The main risk is duplication of records or inconsistent data. That is why it is recommended to: clean up the Excel beforehand, validate each migrated record, and test payroll in parallel for a whole month before turning off the old method.
Can pplPortal's AI make dismissal suggestions?
pplPortal's AI can suggest turnover risks or skills gaps, but it must always be explainable and aligned with the GDPR and AI Act. No dismissal decision can be based on a "black box" — it needs documented human justification.
If I have an old time clock, does pplEvolution work?
It depends on the clock's export format. If it exports to Excel or CSV, pplEvolution integrates. If it is purely analogue (paper cards), you first need to digitalise the data. The initial survey should map exactly which formats you have available.