Performance appraisal on the factory floor: from form to real decision

Sixty-five per cent of employers in Portugal reported in 2024 that they struggled to find professionals with the profile they need — one of the highest figures in the world, according to ManpowerGroup. And yet, most Portuguese factories continue to assess the performance of their people with a paper form or an Excel sheet shared by email, once a year, filled in hastily before the summer holidays.

The problem is not a lack of intention. It is the absence of a connection between what is recorded and what is decided. This article takes a concrete position: performance appraisal in the factory does not fail for lack of tools — it fails because it is designed for HR rather than for the section head, and because appraisal data lives in a silo separate from production, attendance and payroll data. Closing that gap requires an architecture decision, not just a process one.

The form is not the problem. The disconnection is.

Performance appraisal in the factory almost always fails in the same place: between the moment the section head fills in the form and the moment someone makes a decision based on it. That gap can last weeks. Or never close at all.

The form exists. The process exists. What does not exist is the chain that turns the record into an action — a salary progression, a change of position, a training need, a signal of flight risk. When that chain is broken, appraisal becomes an administrative ritual that consumes time without generating any return.

A performance appraisal that feeds no decision is merely bureaucracy with an annual expiry date.

Three pressures converge to make this urgent now. The scarcity of industrial talent — CNC machine operators, knitting technicians, footwear assemblers — has made retention a strategic problem, not just an HR one. The average voluntary turnover in Portugal reached 10.6% in 2023 (Mercer), with 52% of companies admitting they struggle to retain. And the GDPR and Law 58/2019 impose personal data protection requirements that make it unsustainable to keep appraisal records in uncontrolled Excel files, shared by email with no access log.

From annual appraisal to continuous cadence

The annual appraisal model was born in a context where recording was expensive and slow. Today, the cost of recording is close to zero. What is costly is the absence of signal between cycles. A textile factory in the Vale do Ave that only appraises in December does not know, in July, that its best finishing operator is considering leaving — until they have already left.

Continuous cadence does not mean monthly formal appraisals. It means structured micro-records: attendance, productivity per position, disciplinary incidents, completed training, feedback from direct supervisors. These data, when aggregated, build a longitudinal profile that no annual form can capture. The detail that HR manuals rarely mention: in factories with shift rotation, the morning shift supervisor and the night shift supervisor appraise the same operator in completely different ways — and without a system that aggregates both records, the final appraisal reflects only whoever filled in the form last.

The regulatory pressure nobody reads until it is too late

The GDPR (EU Regulation 2016/679), enacted in Portugal by Law 58/2019, classifies performance appraisal data as personal data with specific requirements for legal basis, minimisation and retention. Keeping appraisals in shared Excel, with no access control, no change log and no defined retention policy, is a non-compliance with real consequences — including fines of up to 4% of global turnover.

Law 93/2021, which requires companies with 50 or more employees to maintain a whistleblowing channel, adds another layer: if an employee reports discriminatory practices in appraisal, the company needs to prove that the process was consistent, documented and auditable. An Excel proves nothing. An audit log with timestamps and access profiles does.

The real technical options

There are four approaches in use in Portuguese factories. The visible cost of each is easy to compare. The invisible cost rarely appears in the proposals.

The mistake generalist ERPs systematically make

A generalist ERP models performance appraisal for an office employee: SMART objectives, behavioural competencies, a final score on a scale of 1 to 5. That works for the sales department. It does not work for sewing station 14 on a garment production line in Famalicão.

In an industrial context, appraisal has to map variables that generic forms ignore: multiskilling per position (how many machines the operator masters), actual productivity vs. standard per reference, defect rate per operator, attendance weighted by shift, and behaviour in order peak situations. None of these data live in a form. They live in the production system, the time clock and the CMMS. A generalist ERP does not know those data exist — and it was not designed to go and fetch them.

Appraising a plastic injection operator with the same criteria as an account manager is measuring temperature with a voltmeter.

Trade-offs by dimension

Real cost vs. visible cost

The visible cost of a specialist HR platform is the licence and the implementation. The invisible cost of not having integration is the time spent cross-referencing data manually — an HR technician exporting attendance data from the ERP, cross-referencing it with the appraisals in Excel, calculating bonuses at month end. In factories with 100 to 300 employees, that manual work frequently represents two to three days a month of a qualified person. Multiply by 12 months and compare with the cost of integration. The maths rarely favours Excel.

Integration complexity: what nobody tells you in the proposal

Integration between an appraisal module and operational data requires a well-defined ETL process: which fields flow from the production system into the employee profile, how often, with what transformation rules. Without this mapping done before implementation, the result is empty fields on the appraisal record — and the section head filling in by hand what the system should calculate automatically.

pplPortal, with the pplCore (attendance and payroll) and pplEvolution (appraisal and development) modules, was designed for this context: attendance and productivity data feed directly into the appraisal cycle, with no manual exports. But the integration with the production system — whether KORA Productivity for real-time OEE capture, or the source ERP — has to be planned at the analysis stage, not discovered at the testing stage.

Time to decision: the indicator nobody measures

The true KPI of an appraisal system is not the form completion rate. It is the time between recording an appraisal and the decision it gives rise to. In factories without integration, that time is frequently over 60 days — long enough for a positively appraised employee to receive an offer from another company and accept it. With 65% of Portuguese employers already struggling to recruit (ManpowerGroup, 2024), replacing that employee will cost more than the integration that was not done.

What works in practice

Pattern 1 — Appraisal by position, not by person

The factories that get the most value from the appraisal process first define the position profiles — technical competencies, expected multiskilling, standard productivity indicators — and only then appraise the employee against that profile. This allows consistent comparisons between operators in the same position and identifies training gaps systematically.

In a footwear factory in Felgueiras with 150 employees and 40 distinct positions, this model can reveal that the assembly section has three operators with a profile for progression but with no position available — information that feeds into expansion planning, not just HR. That cross-reference between individual appraisal and organisational capacity is what separates administrative appraisal from strategic appraisal. The People Analytics data that the operations director actually uses takes this form — the article on people analytics in the factory develops the detail.

Pattern 2 — Short cycle with automatic triggers

Instead of a monolithic annual appraisal, the most mature industrial organisations use a 90-day cycle with automatic triggers: when an employee reaches a certain number of months of seniority, when they change position, when they complete a training course, when their attendance index drops below a defined threshold. Each trigger opens a contextualised appraisal form — not the same form for every case.

This model reduces the administrative burden because it spreads appraisals throughout the year, and increases relevance because the appraisal happens at the right moment, not on the right calendar date. For the onboarding of new employees — a critical moment in factories with high turnover — the trigger at 30, 60 and 90 days is particularly effective at detecting integration problems before they become departures. The article on turnover in the factory develops this point in more detail.

Pattern 3 — The section head as the primary user, not HR

The most common implementation mistake in an industrial context is to design the appraisal system for HR and then ask the section head to use it. The section head does not have time, does not have HR training and does not understand why they have to access a portal different from the one they use to manage production.

The projects that work reverse the logic: the section head records observations on the same screen where they manage their team — productivity, incidents, training. HR receives those structured data and completes the formal appraisal. The division of responsibilities is clear: the section head records facts, HR interprets and decides. When this separation does not exist, the form goes unfilled — or is filled in by HR based on nothing.

If the appraisal system forces the section head out of their usual workflow, the completion rate will be proportional to the fear of sanction, not to the quality of the information.

Decision matrix by organisation size

How to link appraisal to decision: five concrete steps

1. Map the positions before the criteria. List all productive positions with the required technical competencies, standard productivity indicators and multiskilling requirements. This map is the reference framework for appraisal — without it, the criteria are generic and the appraisals are inconsistent between section heads.
2. Define the cycle triggers. Identify the events that should open an appraisal: hiring, end of the probationary period, change of position, completion of training, attendance deviation, request for progression. Configure these triggers in the system before launching the first cycle — not after discovering that no one filled anything in.
3. Separate recording from interpretation. The section head records observable facts (productivity, incidents, demonstrated multiskilling). HR interprets and assigns the final score. The system should support this separation with distinct workflows and logs of who recorded what and when.
4. Link appraisal to payroll and training. An appraisal that has no automatic consequence in the system — salary progression, training plan, change of category — does not close the cycle. Configure the business rules that translate the score into action, even if final approval is human. The article on payroll and salary close in the factory details how to automate this link.
5. Measure the time between appraisal and decision. Define an internal KPI: average time between the completion of an appraisal and the decision it gives rise to. Audit quarterly. If that time is over 30 days, the process has a bottleneck — identify where before launching the next cycle.

What appraisal data reveals that other systems do not capture

Flight risk before the departure

The combination of three variables — consecutive positive appraisals without progression, an increase in irregular attendance, and the absence of a training plan — is a flight risk signal that People Analytics systems can detect 60 to 90 days in advance. Without structured appraisal data, that signal does not exist: there is only a surprise at the end of the month when the operator hands in their notice. The article on chronic absenteeism in the factory explores the predictive dimension of this kind of analysis.

Real multiskilling vs. declared multiskilling

In factories with multiple positions and shift rotation, the multiskilling declared in the contract rarely corresponds to the real multiskilling demonstrated. An appraisal system linked to production data can map, position by position, which employees actually operated each machine, with what result and under what conditions. This information is critical for capacity planning — especially during order peak periods, where internal flexibility is the only alternative to subcontracting.

For factories in the clothing sector that subcontract for major international brands, this multiskilling map is also a compliance argument: when the buyer asks who did what and with what qualification, the answer has to be in a system, not in the section head's memory.

The effect of appraisal on absenteeism

The cost of stress and psychological health problems at work in Portugal has been estimated at up to 5.3 billion euros per year, combining absenteeism and presenteeism (Order of Portuguese Psychologists, 2022). A significant part of that cost originates in work contexts perceived as unfair or opaque. Performance appraisal is one of the main vectors of that perception — not because employees want high scores, but because they want the process to be consistent and the decisions that result from it to be visible. When appraisal has no consequence, the message that reaches the shop floor is that effort is not measured. And when effort is not measured, absenteeism ceases to be a health problem and becomes a rational response.

Compliance and auditability: what the GDPR specifically requires

Legal basis and data minimisation

Performance appraisal data are personal data for the purposes of the GDPR. The most common legal basis in an employment context is legitimate interest or the performance of the employment contract — but it has to be documented in the company's record of processing activities (RoPA). Each field of the appraisal form has to have a justification of necessity: if it does not feed any decision, it should not exist. A form with 40 fields, of which 30 are never read, is not merely inefficient — it is an active non-compliance.

Retention and erasure

Define the retention period for appraisal data before implementing the system. Common practice in Portugal sits between 3 and 5 years after the termination of the employment contract, aligned with labour limitation periods. The system has to support automatic erasure or anonymisation at the end of that period — something an Excel does not do, and which no one remembers to do manually.

Access and audit log

Who can see whose appraisal? The section head sees their direct team. HR sees the whole organisation. The employee has the right of access to their own data (Article 15 of the GDPR). The system has to support these access profiles and keep a log of who accessed, when and with what changes. Without this log, the company cannot respond to an audit request — neither from the CNPD, nor from a labour court.

For organisations with integrated document management processes, Document Management with legal archiving and an approval workflow complements the appraisal module — especially for archiving historical appraisals with legal value.

The role of BI in appraisal: from record to decision dashboard

Appraisal data in isolation have limited value. Cross-referenced with production, attendance, training and turnover data, they become a management instrument. A Qlik Sense dashboard that shows, per section, the distribution of appraisal scores, the turnover rate, the average OEE and the training cost per employee gives the operations director a view that no traditional HR report provides.

The technical detail that rarely appears in the proposals: for this dashboard to work, the appraisal data have to be in a Data Warehouse shared with the operational data. If HR uses a system separate from the ERP and production, integration requires a well-defined ETL process — and someone responsible for keeping it up to date when the forms change. The article on self-service BI in the factory develops this point with practical examples.

Appraisal as a system, not as an event

Performance appraisal in the factory only generates value when it is treated as a continuous system — with structured input data, business rules that translate records into decisions, and a feedback loop that closes on the employee. The form is merely the visible interface of a process that begins in production data and ends on the payslip, in the training plan or in the progression decision.

Factories that still treat appraisal as an annual compliance event are wasting the one asset that automation does not mind replacing: the operational knowledge of their best employees. When those employees leave — and with 10.6% average voluntary turnover in Portugal, some will leave — the company discovers that it had recorded nothing worth transferring. What stayed was the Excel. What left was the operator who knew how to make the machine run out of spec without stopping the line.

For the broader context of HR digitalisation in an industrial environment, the article on rotating shift schedules and the article on people analytics in the factory complete the operational picture within which performance appraisal sits.

Sources

• ManpowerGroup, Talent Shortage Survey 2024 — talent scarcity in Portugal (65% of employers struggling to recruit).
• Mercer, Global Talent Trends 2023 — average voluntary turnover in Portugal of 10.6% and retention difficulty in 52% of companies.
• Order of Portuguese Psychologists, Psychological Health at Work Report 2022 — estimated cost of stress and absenteeism/presenteeism of up to 5.3 billion euros per year in Portugal.
• Regulation (EU) 2016/679 (GDPR) and Law 58/2019 — protection of personal data, including performance appraisal data in an employment context.
• Law 93/2021 — regime for the protection of whistleblowers and the mandatory whistleblowing channel for organisations with 50 or more workers.

Frequently asked questions

Why do most Portuguese factories still use paper forms to appraise performance?

The article points out that the problem is not a lack of intention, but the absence of a connection between the record and the real decision. Factories keep paper forms or Excel because appraisal was historically designed for the HR department, not for the section head. Without an architecture that integrates these data with production and attendance, the form becomes an annual administrative ritual with no practical return.

What is the critical gap where performance appraisal fails?

The failure occurs between the moment the section head fills in the form and the moment someone makes a decision based on it. That gap can last weeks or never close at all. Without a chain that turns the record into a concrete action — salary progression, change of position, training or identification of flight risk — appraisal remains bureaucracy with no value.

What pressures make it urgent to reform appraisal in the factory?

Three pressures converge: the scarcity of industrial talent that makes retention strategic; the high voluntary turnover, with companies admitting they struggle to retain employees; and the requirements of the GDPR and Law 58/2019 that make it unsustainable to keep appraisal records in uncontrolled Excel files, shared by email with no access log.

What does moving from annual appraisal to continuous cadence mean?

It does not mean monthly formal appraisals, but structured micro-records: attendance, productivity per position, disciplinary incidents, completed training and supervisor feedback. These aggregated data build a longitudinal profile that no annual form can capture, making it possible to identify signals of risk — such as an operator considering leaving — between appraisal cycles.

What legal risks exist in keeping appraisals in shared Excel?

The GDPR and Law 58/2019 classify appraisal data as personal data with specific requirements for legal basis, minimisation and retention. Keeping appraisals in Excel with no access control, no change log and no defined retention policy constitutes non-compliance with real consequences, including fines of up to 4% of global turnover.

Why does a generalist ERP not solve the problem of appraisal in the factory?

A generalist ERP models appraisal for office employees with SMART objectives and behavioural competencies. It does not map real industrial variables: multiskilling per position, productivity vs. standard, defect rate, attendance weighted by shift or behaviour in order peaks. The generic forms ignore the specificity of the factory context.

What is the main risk of a standalone specialist HR platform?

The main risk is the creation of silos: HR and appraisal data remain separated from production, operations and payroll data. Without integration between these systems, appraisal does not feed real operational decisions and remains disconnected from the factory's productive reality.