Performance management trends for 2027: six shifts HR should plan for

Managers are already using general-purpose assistants to draft reviews, whether or not HR has a policy. Our prediction for 2027 is that this shadow use gets pulled into the open. HR teams will choose approved tools, decide which employee data may go into them and write down where a person must take over. The organisations that ban AI outright will find it used anyway, with less control.

The EU AI Act is the reason this happens faster in Europe. It lists AI used to monitor and evaluate the performance and behaviour of people in work relationships among its high-risk uses, alongside AI used for decisions on promotion and termination. That brings obligations for the employer using the system, not only for the vendor that builds it.

Job titles are a poor unit for comparing people. Two “senior analysts” in different departments can do very different work, and two people with different titles can do work of equal value. We expect 2027 to be the year more organisations assess against defined skills and competencies, each with levels described in behaviour, mapped to what each role requires.

Three pressures drive this at once. Pay transparency asks employers to compare work of equal value, which needs criteria that travel across teams. Internal mobility and succession need to know who can do what before the vacancy appears. And AI-assisted drafting only produces something useful when there is a defined standard to write against.

The trap is building a skills taxonomy so large nobody maintains it. The organisations that get this right start small: a short, shared library, a handful of required competencies per role and assessment evidence that keeps the inventory current. A matrix updated by the review cycle beats a perfect one updated by nobody.

For a decade, a vocal group of employers dropped performance ratings in favour of conversations. Many found that the judgement did not disappear; it moved somewhere less visible, such as a spreadsheet used to set pay, a talent review or a manager’s memory. Our view is that 2027 settles the argument in favour of neither side.

Where performance affects pay, bonus or promotion, the Pay Transparency Directive makes an unexplained judgement a liability, whether or not it is called a rating. So the realistic direction is not “no ratings” but fewer, better-defined ones: a short scale with every level anchored to observable behaviour and outcomes, evidence attached to each rating and calibration before anything is final. Where performance does not drive money, such as in development-only check-ins, a rating-free conversation remains a good choice.

Directive (EU) 2023/970 gave member states until 7 June 2026 to bring it into national law. Ireland confirmed it would miss that deadline and has signalled a phased implementation, so Irish HR teams go into 2027 with the direction clear and some of the detail still moving. Under the Directive, the first gender pay gap reports from employers with 250 or more workers are due by 7 June 2027.

The part that reaches the review cycle is quieter than the reporting. Employers must make available the criteria used to set pay and pay progression, and those criteria must be objective and gender-neutral. In most organisations, ratings feed merit increases, bonuses and promotion, which makes the review criteria part of the pay criteria.

Our prediction: in 2027, HR teams will audit their review process the way finance teams audit a ledger. Can we show the criteria, the evidence behind each rating, how the scale was applied across comparable roles and why any rating changed? Teams that can answer those questions will find the Directive manageable. Teams that cannot will find their review cycle is the weakest link in explaining a pay difference.

Continuous feedback won the argument years ago; few HR teams still defend a once-a-year conversation as the only one. What has lagged is the practice. Regular check-ins only work if managers know how to run them, write down what they observe and give feedback that is specific enough to act on. Without that, a continuous process becomes an annual review with more meetings.

We expect the investment in 2027 to shift from process design to manager capability. That means short, practical training before each cycle rather than a one-off workshop; templates and question banks that make a good check-in easy; and lighter administration, which is where AI drafting genuinely helps, so the time saved goes into the conversation. Every other trend in this article lands on managers: they apply the skills framework, own the AI-assisted draft and defend the rating that feeds pay.

Ratings, feedback, development notes and 360 comments are personal data, and employees know how to ask for them. As AI and more frequent feedback produce more performance data, we expect privacy to move from the end of a software evaluation to the beginning.

CLEAR Talent is performance management software built around the direction this article describes. Ava, the AI assistant inside it, drafts review narratives from the goal scores, competency ratings, 360° feedback and check-in notes already in the platform. Every output is labelled, editable and auditable, managers accept, amend or reject it, and every edit is logged. Ava never makes the final rating; she flags rating patterns that may indicate gender, recency or leniency bias before a manager submits. Ava runs on isolated tenants, your employees’ data is never used to train models for other customers, and AI can be switched off per cycle, team or module.

Competencies sit in one library with behavioural levels and are mapped to the levels each role requires, so the skills inventory is built from the assessments you already run. Calibration views flag inconsistent ratings across managers, and DSAR exports, erasure workflows and a data processing agreement cover the privacy side. CLEAR Talent does not certify your compliance with the EU AI Act, and it is not a pay or pay-gap reporting system.