Should performance reviews be linked to pay? Usually yes, but indirectly. Linking reviews to pay means the outcome of a review (a rating, a grade or a calibrated position against a talent bar) is one of the inputs to salary progression or bonus, alongside market position and budget. The evidence supports making that link visible and consistent; it does not support running the development conversation and the pay decision in the same meeting, or turning a rating into a raise by formula without a check for bias. The CIPD’s factsheet Performance-related pay (November 2025) puts the order plainly: define, measure, appraise and manage performance first, and only then use pay to reward it. For a company of 100 to 1,500 people, the real question is not whether to link, but how tightly, and what has to be in place before the first merit round depends on review data.
Key takeaways
- Link the decision, split the conversation. A 1965 General Electric study by Meyer, Kay and French found criticism in appraisal interviews made people defensive, and recommended holding separate discussions for development and for salary.
- Money does move performance, mostly quantity. Cerasoli, Nicklin and Ford’s 2014 meta-analysis (183 studies, 212,468 people) found incentives better predicted quantity of performance, while intrinsic motivation better predicted quality.
- Team-based rewards are worth considering. Garbers and Konradt (2014) found a positive effect for individual financial incentives (g = 0.32) and a larger one for team-based rewards (g = 0.45), stronger when shared equitably rather than equally.
- The pay step is where bias hides. Castilla (2008) found women and minority employees received smaller salary increases than white men with the same performance scores, in the same job and unit, under the same supervisor.
- Transparency is becoming a legal duty in the EU. The EU Pay Transparency Directive (2023/970) requires employers to make the criteria for pay and pay progression easily accessible, and those criteria must be objective and gender neutral.
Follow us on Google
What “linked to pay” actually means
The phrase covers several quite different designs, and most arguments about it are really arguments about which one is meant. The CIPD factsheet describes two common shapes for a pay review under performance-related pay: “all-merit”, where the whole increase depends on the individual rating, and “basic-plus-merit”, where everyone receives a general increase and a further element depends on performance. It also notes that individual performance-related pay is widespread in the private sector and close to the norm in parts of financial services, while most public-sector employers give a basic percentage increase.
There is also a third choice that often goes unstated: whether the link is a formula (rating 4 equals a 5% rise) or an input to a judgement (rating, position in range and market data go to a pay review, which decides). That choice matters more than the headline. A formula is transparent but rigid and magnifies any error in the rating. A judgement-based link is flexible but needs to be explained, or employees will assume there is no link at all. The performance management framework on this site treats pay as one of four outcomes of a calibrated grade, alongside promotion, development and an improvement plan.
What the evidence says
Pay motivates, but not in the way people fear or hope. The long-running worry is that money “crowds out” intrinsic motivation. Cerasoli, Nicklin and Ford’s 40-year meta-analysis in Psychological Bulletin (2014) found the two are not necessarily antagonistic. Intrinsic motivation predicted performance whether or not incentives were present. Where incentives were directly tied to performance, intrinsic motivation mattered less; where they were indirectly tied, it mattered more. Incentives were the better predictor of how much work got done, intrinsic motivation of how good it was. For knowledge work, where quality is most of the job, that argues for an indirect link rather than a piece-rate.
Financial incentives have a real, moderate effect. Garbers and Konradt’s meta-analysis of 146 studies (Journal of Occupational and Organizational Psychology, 2014) found a positive overall effect for individual incentives (g = 0.32), smaller for less complex tasks (g = 0.19), and a larger effect for team-based rewards (g = 0.45). Team rewards worked better when distributed equitably, by contribution, than equally.
The review meeting is the wrong place to discuss money. Meyer, Kay and French’s Split roles in performance appraisal (Harvard Business Review, 1965) studied appraisal interviews at General Electric. Their conclusions still read as current: criticism tended to produce defensiveness rather than improvement, goal-setting worked better than criticism, coaching should be day-to-day, and appraisals should be split by purpose, with salary handled separately from development.
Accurate ratings do not guarantee fair pay. Emilio Castilla’s study of personnel data in a large service organisation, Gender, race, and meritocracy in organizational careers (American Journal of Sociology, 2008), found no gap in ratings once job and work unit were controlled for, yet women and minority employees received smaller salary increases than white men with the same scores. He calls this performance-reward bias and points to limited transparency and accountability in the step from rating to pay as the cause.
Three ways to connect reviews and pay
| Model | How it works | Strength | Main risk | Usually fits |
|---|---|---|---|---|
| Direct formula | Calibrated rating maps to a fixed merit percentage or bonus multiplier | Simple to explain; quick to run | Any rating error or bias flows straight into pay; managers inflate ratings to protect raises | Roles with countable output; mature, calibrated rating systems |
| Informed judgement | Rating is one input with position in range, market data and budget; a pay review decides | Flexible; separates “how well” from “what we can afford” | Feels arbitrary unless the criteria and rationale are written down and shared | Most 100 to 1,500-person companies with mixed roles |
| Decoupled | Pay moves on market and role; reviews drive development and promotion only | Keeps review conversations about growth | High performers see no link; pay drifts away from contribution over time | Organisations with structured pay scales or banded progression |
In all three, the separate-conversations finding still applies: the development review happens first, and the pay outcome is communicated later, once calibration and the pay review are done.
When a 100 to 1,500-person company should link them
Link reviews to pay when three conditions hold. First, ratings are calibrated across managers, so a “strong” in one team means the same in another; the calibration meeting guide covers how to run that step. Second, the rating scale is narrow enough to defend; the rating scales guide compares the common options. Third, someone checks pay outcomes by group before they go out.
If those are not in place, a tight link will reward the most generous managers rather than the strongest performers. In that case, start with basic-plus-merit and a modest merit pot while the rating system matures.
Employers in Ireland or elsewhere in the EU have an extra reason to tidy this up. Article 6 of Directive (EU) 2023/970 requires employers to make easily accessible the criteria used to determine pay, pay levels and pay progression, and those criteria must be objective and gender neutral; member states may exempt employers with fewer than 50 workers from the pay-progression part. Member states had to transpose the Directive by 7 June 2026. Article 9 adds gender pay gap reporting, including on variable pay: from 7 June 2027 for employers with 150 or more workers, and from 7 June 2031 for those with 100 to 149. Check how the transposing law in each country where you employ people applies to you. In the UK, equal pay for equal work is already required under the Equality Act 2010, so a pay step you cannot explain is a risk in both jurisdictions.
Where the link goes wrong
- Rating inflation. When a rating sets the raise, managers protect their teams by rating up. Calibration and a clear talent bar are the counterweight.
- One meeting for two jobs. Telling someone their number and their raise in the same conversation crowds out any discussion of how to improve, which is what Meyer, Kay and French observed.
- An unexplained final step. The adjustment between recommended and approved increase is where Castilla located the bias. Record who changed what and why.
- Pay for things the review did not measure. If behaviours and values count in the review but not in pay, or the reverse, people learn which part really matters. Keep the same scorecard for both; see performance review bias for the patterns to watch.
- Individual-only rewards for team work. Garbers and Konradt’s results suggest a team element is worth testing where outcomes are shared.
Our perspective
When People leaders ask me this question, the link itself is rarely the problem. The problem is that nobody can show the chain from evidence to rating to raise. Ratings come back late from a form, the merit spreadsheet is built from a copy of them, and the adjustments made in the final week live in someone’s inbox. Employees then hear about pay and performance as two unrelated events.
My advice is to make the chain visible before you make it tight. Calibrate first, write the pay criteria down in a paragraph anyone can read, and keep a one-line rationale for every adjustment. Then run an outcome check by gender and level before letters go out. If you can do that comfortably, you can afford a firmer link. If you cannot, a firmer link will only make the gaps more expensive.
— Ash Taherian, Product Sales Executive, Revolut People
How Revolut People supports the link
I work at Revolut People, so weigh this section accordingly. The part of the product that matters here is reviews with calibration: one scorecard per role, a calibration step before ratings are final, and a single record of each outcome and its rationale. That record is the evidence a pay review needs, and it is what lets you explain afterwards why two people with the same rating were treated differently. Revolut’s own story describes a compensation overhaul as one of the first projects once leadership had current performance data (Revolut customer story). The limits are real: software will not write your reward philosophy, set your merit budget or make the differentiation call for you. For how this looks from a People leader’s seat, see Revolut People for CHROs.
If you want a second view on whether your review data could carry your next pay round, book a 15-minute review. Bring your rating scale and last year’s merit matrix with names removed.
Step plan: one pay cycle
- Eight weeks before the pay review: choose the model. Decide between formula, informed judgement and decoupled, and whether there is a team element. Write it in one paragraph.
- Seven weeks before: publish the criteria. Share what inputs determine pay progression (rating, position in range, market data, budget) in plain language. For EU staff, this is the Article 6 duty.
- Six to four weeks before: run reviews as development conversations. No numbers on pay in the review meeting. Use the review-cycle checklist to keep the timeline.
- Three weeks before: calibrate. Hold calibration by department, then a cross-department check at the top and bottom of the distribution. The calibration meeting template gives an agenda.
- Two weeks before: run the pay review and log every adjustment. Record recommended and approved increases side by side with a one-line reason for any difference. If you use a formula or informed-judgement model, a merit matrix is the usual way to turn a calibrated rating and position in range into that recommendation.
- One week before: check outcomes by group. Compare average increase by rating band across gender and level. Investigate any gap among people with the same rating before letters go out, then hold separate pay conversations.
Bottom line
Link reviews to pay, but through a calibrated rating and a written set of criteria, not through the review meeting itself. The research supports a moderate, visible link and a separate conversation, and it is clear that the riskiest moment is the step from rating to raise. For a quick baseline on whether your ratings are ready to carry that weight, run the free performance health check or the review bias checker. If you would like a second view of your set-up, send a short brief through the People Performance Playbook intake.
FAQ
Should the pay conversation happen in the same meeting as the performance review?
It is better to separate them. The 1965 General Electric study by Meyer, Kay and French recommended split discussions, because salary in the balance makes people defensive and crowds out the development discussion. Many companies hold the review first and communicate pay once calibration and the pay review are complete.
Why do companies calibrate performance ratings before pay decisions?
Because managers rate to different standards. Calibration compares ratings across teams against the same scorecard before they are final, so that the same rating means the same thing when it reaches a pay review. Without it, pay tends to follow manager generosity rather than performance.
How many points should a rating scale have if it drives pay?
Few enough that managers can tell the points apart and defend them. Many companies use four or five points; what matters more is a written definition for each point and calibration against it. The rating scales guide on this site compares the options.
Does the EU Pay Transparency Directive apply to UK companies?
It is EU law, so it does not apply to employment in the UK in its own right. A UK company that employs people in Ireland or another EU member state should check how that country’s transposing law applies to those staff. UK employers are already bound by the equal pay provisions of the Equality Act 2010.
Sources
- CIPD: Performance-related pay, factsheet (3 November 2025) - all-merit and basic-plus-merit awards; prevalence in private and public sectors; define, measure, appraise and manage performance before rewarding it
- Herbert H. Meyer, Emanuel Kay and John R. P. French Jr.: Split roles in performance appraisal, Harvard Business Review (January 1965) - General Electric study of 92 employees; separate appraisals by purpose
- Christopher P. Cerasoli, Jessica M. Nicklin and Michael T. Ford: Intrinsic motivation and extrinsic incentives jointly predict performance: a 40-year meta-analysis, Psychological Bulletin 140(4), 980-1008 (2014) - k = 183, N = 212,468; incentives predict quantity, intrinsic motivation predicts quality
- Yvonne Garbers and Udo Konradt: The effect of financial incentives on performance, Journal of Occupational and Organizational Psychology 87(1), 102-137 (2014) - 146 studies, n = 31,861; individual g = 0.32, team g = 0.45
- Emilio J. Castilla: Gender, race, and meritocracy in organizational careers, American Journal of Sociology 113(6) (2008) - performance-reward bias in salary increases at equal performance scores
- European Parliament and Council: Directive (EU) 2023/970 on pay transparency (May 2023) - Article 6 pay-progression criteria; Article 9 reporting thresholds; Article 34 transposition by 7 June 2026
- UK Parliament: Equality Act 2010, Part 5, Chapter 3: Equality of terms (2010)
Related reading
- Performance management framework
- Gartner on pay for performance: six fixes so review outcomes can carry pay decisions
- UK and Ireland performance review benchmarks 2026
- How to use a merit matrix to link ratings to pay
- How to run a performance calibration meeting
- What is continuous performance management?