Separating the job evaluation method from the software that runs it

An organisation looking for job evaluation software is usually making two decisions at once. The first is how the relative value of its jobs will be determined. The second is how that determination will be carried out, reviewed, recorded and governed across the organisation.

Treating them as a single decision creates two risks. A scheme can end up being judged on the quality of its interface. A weak method can also look more defensible once it sits inside a well-designed system.

It helps to separate job evaluation into three layers. The methodology determines what is evaluated and how relative value is decided. The process determines how people apply that methodology. The technology determines how that process is enabled, controlled and evidenced. This article looks at what research, regulation and case law say about each layer.

The methodology

The methodology is the framework that determines relative job value. It defines the factors or dimensions being assessed, what each one means, the levels within each factor, how levels are scored or weighted and how total scores translate into grades or bands.

This is where the substantive decisions about value are made. The CIPD’s factsheet on job evaluation and market pricing treats a scheme as something an organisation chooses and then implements, starting with the choice between analytical and non-analytical approaches. It also makes the point that job evaluation needs regular review as jobs, work and business needs change.

Legislation frames it in the same terms. Section 80(5) of the Equality Act 2010 defines a job evaluation study as one undertaken to evaluate jobs “in terms of the demands made on a person”, by reference to factors such as effort, skill and decision-making. Article 4(4) of the EU Pay Transparency Directive (2023/970) requires pay structures to allow the value of work to be assessed against objective, gender-neutral criteria including skills, effort, responsibility and working conditions. Member states were required to transpose the Directive by 7 June 2026. Both pieces of legislation set out the basis on which work should be valued without specifying any technology for applying it.

The EU-wide guidelines on gender-neutral job evaluation and classification, updated by the European Commission and the European Institute for Gender Equality in March 2026, take the same approach. They offer a step-by-step toolkit for organisations of different sizes and are guidance rather than a legal act.

The quality of the method has to come first. Nan Weiner’s 1991 critique of job evaluation systems in Human Resource Management Review identified four recurring problems: inconsistent construct formation, overlap between factors, level statements anchored to the existing hierarchy and gender bias. Her standard for a good scheme was that each subfactor defines a clear construct, measures it consistently at every level and is measured only once.

She also set out how gender bias enters a scheme’s design. One route is valuing a factor in jobs traditionally held by men but not in jobs traditionally held by women. Another is leaving out factors found in women’s jobs altogether.

UK equal pay law has tested the consequences of a gap in a scheme’s design. In Element and others v Tesco Stores Ltd [2022] EAT 165, the Employment Appeal Tribunal upheld a finding that a 2014 scoring exercise was not a job evaluation study under section 80(5). One reason was that it did not cover the demands made on the job holders. The original tribunal noted in particular that the exercise had no factor for the physical effort or skills the jobs clearly required.

Software applies the factors it is given. A factor missing from the scheme will be missing from every evaluation the software processes, however well the surrounding workflow is designed.

The process

The process is how people apply the methodology in practice. It covers how job information is gathered, who evaluates, how results are checked for consistency, who reviews and approves them, how disagreement is raised and resolved, what is recorded and when a job is looked at again.

This layer controls whether a sound method produces consistent results. In a controlled study published in the Journal of Applied Psychology in 1988, Hahn and Dipboye asked participants to evaluate 23 benchmark jobs on 10 dimensions under different conditions of training and job information. Participants given both a job title and a job description were generally more accurate and reliable than those given a title alone. Training mainly changed how ratings were distributed, reducing leniency. The most accurate and reliable evaluators tended to be those who had both training and full job information.

Every participant applied the same evaluation dimensions. The differences in accuracy and reliability came from the information and preparation they were given. In an organisation those conditions are set by the process: what is collected about each job, who evaluates it, how evaluators are prepared and how results are checked. Two organisations using the same scheme can reach different results for similar jobs when these conditions differ. The same variation can occur within one organisation, between departments, sites or individual evaluators.

Completion and agreement are also part of the process. The tribunal’s other reason in the Tesco case was that the exercise was still a work in progress. The Pay Transparency Directive requires the criteria used to assess the value of work to be agreed with workers’ representatives where such representatives exist. Both requirements concern how an evaluation is carried out and agreed rather than what the criteria measure.

The technology

The technology is the digital infrastructure that makes the process workable at scale. It includes user roles and permissions, workflow stages and status tracking, evidence capture, version control, an append-only audit trail, role libraries, notifications, reporting and integration with HR systems.

Computer support for job evaluation is well established, with published research on job evaluation expert systems dating back to at least 1995. Michael Armstrong’s Job Evaluation Handbook (2018) describes computer-aided schemes as generally built on a conventional point-factor scheme. The software either converts job information into factor scores using predetermined rules or takes the job holder and their manager through a series of linked questions that lead to a score. In both forms the rules come from the method.

Computerising the calculation and digitising the governance of job evaluation are different undertakings. A spreadsheet can calculate a score. An online form can replace a paper questionnaire. A database can store results. A platform that supports the process has to manage the full lifecycle of a decision: creation, evidence, evaluation, review, challenge, approval, recording, reporting, monitoring and re-evaluation.

The need to govern that lifecycle is greatest where an organisation has to evidence its decisions. Under Article 7 of the Pay Transparency Directive, workers can request information on their own pay level and on average pay levels, broken down by sex, for workers doing the same work or work of equal value. Article 18 places the burden of proof on the employer in pay discrimination proceedings where it has not implemented the Directive’s pay transparency obligations. Article 21 requires limitation periods for equal pay claims of no less than three years.

An employer may therefore need to explain an evaluation years after it was made. That means showing who carried it out, the information used, the version of the criteria applied and the review that took place. Recording those details reliably is where most of the value in the technology layer sits.

One architecture, more than one method

Job evaluation is a family of methods rather than a single one. A scoping review of 199 academic articles, published in Compensation & Benefits Review in 2025, found that research moved from methodological questions in the 1940s towards adapting job evaluation methods to specific contexts from 2000 onwards.

Separating the layers has a practical consequence for how job evaluation software is designed. The workflow and governance infrastructure can be built independently of any particular set of factors. The factors, questions, levels, scoring rules and grade boundaries can then be held as configuration and versioned separately from the workflow. Designed this way, the same underlying architecture can support different evaluation methodologies where their requirements are compatible.

That qualification matters. A whole-job ranking exercise, a point-factor scheme and a levelling framework ask for different information and different kinds of judgement. Some methods also carry their own requirements for panel composition or approval. The claim that holds up is more limited: the parts of a system that manage people, stages, evidence and records do not necessarily need to be rebuilt each time the evaluation logic changes.

The separation is useful for three reasons.

First, methods change. The CIPD advises regular review as jobs and business needs change. The Pay Transparency Directive gives employers in EU member states a specific reason to revisit their criteria. The 2025 scoping review found hardly any research on changes to job evaluation methods and criteria. That leaves employers with a practical question that has had little research attention: how should changes to evaluation criteria be managed and evidenced over time? A system that hard-codes the method turns every change to the method into a software change.

Second, results need to remain explainable after a method changes. Each evaluation should record the version of the method that produced it. An organisation can then show which criteria applied to a decision made several years earlier, even after those criteria have been revised.

Third, some organisations run more than one scheme, for example after a merger or for different parts of the workforce. A shared workflow allows one governance process to apply across schemes while each scheme keeps its own evaluation logic.

Questions to ask of any job evaluation system

• Which parts of the method can be configured and who controls changes to them?

• Does the system distinguish a change to the method from a change to the workflow?

• Does every evaluation record the version of the method and the job information it was based on?

• Can reviewers see the evidence behind each factor level as well as the total score?

• How are disagreements, appeals and re-evaluations recorded?

• Is the audit trail append-only?

• What reporting shows consistency across evaluators, departments and sites?

• Can the organisation introduce a revised method while preserving the history of evaluations made under previous versions?

Conclusion

At organisational scale, job evaluation is a governed decision-making process as much as a scoring exercise. The method determines what is valued. The process determines how that valuation is carried out. The technology determines how the result is controlled, evidenced and maintained over time.

Assessing each layer on its own terms makes it easier to see where a weakness sits: in the criteria, in how they are applied or in the record of what was decided.

I would be interested to hear whether others assess the method and the system separately when reviewing their own job evaluation arrangements.

Sources