Performance Review Statistics 2026: Only 22% Call the Process Fair, and 2% of CHROs Think It Works
The performance review is the most universally administered ritual in corporate life and the most universally disliked. That much is folklore. What is less well known is that the disapproval is not a bottom-up complaint: it is strongest at the top. The chief human resources officers who own these systems rate them lower than the employees who sit through them.
This page assembles the verified 2026 dataset in one place — who believes the process works, how often it actually happens, what evaluations are built from, how biased the written feedback is, where AI has arrived in the last eighteen months, and which interventions have a published effect size. Every figure carries its source, reference period and, where published, its sample size, because this topic has an unusually bad citation hygiene problem: several of the most-quoted "performance review statistics" circulating in 2026 are from 2015 fieldwork, and at least one cannot be traced to any published report at all. Those are listed in section 10 rather than repeated as fact.
1. The employee verdict, measured properly
The largest recent employee-side dataset on this question is Gallup's U.S. Workplace Panel Study: 18,665 US employees aged 18 and over, working full- or part-time, surveyed 9–24 August 2023, with a margin of error of ±1.1 percentage points. That sample size matters here, because most performance-review statistics in circulation come from vendor panels of one to three thousand people.
Gallup's scale is deliberately demanding. "Strongly agree" is the top box of a five-point scale, and Gallup's research convention is that only the top box predicts behaviour. So these are not satisfaction scores; they are counts of people who are unambiguously positive.
| What US employees say about performance management | Strongly agree | Source |
|---|---|---|
| My performance review process is fair and transparent | 22% | Gallup panel, Aug 2023 |
| Reviews are transparent, fair, or inspire better performance (any of the three) | 20% | Gallup panel, Aug 2023 |
| I know what is expected of me at work | 47% | Gallup panel, Aug 2023 |
| My performance is managed in a way that motivates outstanding work | 20% | Gallup, performance management series |
| My manager involves me in setting my goals | 30% | Gallup, performance management series |
| The performance metrics I am held to are within my control | 21% | Gallup, performance management series |
| My performance review inspires me to improve | 14% | Gallup, performance management series |
The "I know what is expected of me" line is the one to watch, because Gallup has tracked it for a decade and it is moving in one direction. It stood at 61% in 2015, 56% immediately before the pandemic, and 47% in the August 2023 panel. Role clarity is the foundation every other performance measure sits on, and it has lost fourteen points in eight years.
Fewer than half of American employees can strongly agree they know what is expected of them at work. Every other number on this page is downstream of that one.
2. Nobody in the chain believes in it — least of all the people who own it
The usual framing is that employees resent reviews and HR defends them. The data says the opposite. In June and July 2023, Gallup surveyed 135 chief human resources officers at global Fortune 500 companies. Two percent strongly agreed that their performance management system inspires employees to improve.
That figure is not an outlier. Mercer's Global Performance Management Study found that 2% of companies believed their performance management system delivered exceptional value — a different population, a different question, four years earlier, and the same number.
Who says the performance management process works
Percentage expressing confidence, by population. Sources: Gallup (2023), Deloitte (2025), WTW (2025).
Deloitte's 2025 Global Human Capital Trends survey — nearly 10,000 business and HR leaders across 93 countries — produced the two figures that matter most here. 72% of workers and 61% of managers could not say they trust their organisation's performance management process. Invert them and you get the people who can: 28% of workers, 39% of managers.
Two further Deloitte findings explain why confidence is so low even among the people running the process. 75% of organisations rate their own ability to accurately evaluate an individual worker's value as "not very" or "not at all" effective. And only 6% both use data and evidence well and enhance worker trust — the two things a performance system is supposed to do simultaneously.
Betterworks' survey data, cited in the same Deloitte analysis, puts the employee verdict in blunter language: 64% of workers see performance reviews as a complete waste of time that does not help them perform better, and fewer than one in three think reviews are very fair and equitable.
3. How often reviews actually happen — and what changes when they happen more
Ten years of "kill the annual review" commentary has produced less change than the commentary implies. In Gallup's August 2023 panel, 56% of employees formally reviewed their performance goals with their manager once a year or less. In Gallup's earlier performance management series the split was 48% annually and 26% less than once a year — leaving roughly a quarter of the workforce with no annual conversation at all.
Frequency is also the one variable in this dataset with a clean, published effect on the outcome everyone complains about. Employees who have quarterly progress checks are:
- 90% more likely to be engaged than employees who do not; and
- 2.1× as likely to say the performance review process is fair and transparent.
Applied to the 22% baseline, a quarterly cadence takes perceived fairness to about 46% — a 24-point move, achieved without redesigning a single form, rating scale or competency framework.
4. What performance evaluations are actually built from
Before asking whether reviews are fair, it is worth knowing what they are made of. Gallup's 2023 panel asked employees directly.
What US employees say their evaluation is based on
Percentage of employees reporting each input. Source: Gallup U.S. Workplace Panel Study, August 2023, n = 18,665.
Two things stand out. The dominant input is manager observation — a subjective instrument, which is precisely why section 5 on language bias matters. And the customer, the party whose opinion arguably decides whether the work was any good, appears in 19% of evaluations.
On the mechanics of rating, WTW's September 2025 research found 45% of participating organisations still using a five-point rating scale, while 54% have changed or are considering changing their scale. The category that was declared dead in 2015 is, in practice, being adjusted rather than abandoned.
| Organisation-side practice (WTW, September 2025) | Share |
|---|---|
| Say their performance management process is effective at meeting employee expectations | 39% |
| Say managers are effective at providing coaching and feedback | 20% |
| Believe productivity would rise by 10% or more if the process were optimised | ~50% |
| Use a five-point rating scale | 45% |
| Have changed, or are considering changing, their rating scale | 54% |
| Have integrated skills into performance management | 54% |
| Use pay-for-performance to reward high performers | 68% |
| Use pay-for-performance to motivate and engage employees | 53% |
| Currently use AI somewhere in performance management | 37% |
The 39% and the 20% belong together. Only two in five organisations think the process delivers, and only one in five think their managers can coach — which locates the failure in capability rather than in software. That is the same conclusion reached from a different direction in our manager effectiveness dataset.
5. The feedback itself: what a 23,000-review corpus shows
Opinion surveys tell you how reviews feel. Textio's work is different in kind: it analyses the actual text of performance feedback. Its 2024 report examined 23,000 performance reviews and surveyed 450 people; its 2022 predecessor analysed roughly 25,000 reviews. The findings are consistent across both corpora, and they are uncomfortable.
The headline result is counter-intuitive. High performers — up to about 15% of a workforce — receive 1.5× more feedback than everyone else, and the lowest-quality feedback of any group. More words, less substance. Textio attributes 10% of all attrition to low-quality feedback, and found that people who receive it are 63% more likely to leave within a year.
| Measured language bias in performance feedback | Finding |
|---|---|
| Feedback about personality rather than work, women vs men | 22% more for women |
| Unactionable feedback, women vs men | 2× as much for women |
| Problematic feedback among high performers, women vs men | 38% more for women |
| Unactionable feedback, Black employees vs white or Asian colleagues | 2× as much |
| Recall being described as "emotional" at work | 78% of women vs 11% of men |
| Recall being described as "likable" at work | 19% of women vs 71% of men |
| Recall negative "emotional" characterisations, by group | >60% of Hispanic and Black employees vs 21% of white employees |
| Positive stereotyping as "intelligent" | White and Asian employees 2× as likely as Hispanic and Black employees |
The recall figures produce a ratio worth stating plainly: women are about seven times as likely as men to remember being called emotional at work (78% against 11%), while men are close to four times as likely to remember being called likable (71% against 19%). These are self-reported recollections rather than counts from the review corpus — a weaker instrument than the text analysis, and worth labelling as such when citing.
The dominant input to a performance evaluation is manager observation. The dominant defect in the written record is that it describes who someone is rather than what they did.
The practical reading is that "more feedback" is not the goal and never was. High performers already get the most words. What they do not get is anything they can act on — which is why the interventions that work in section 8 are about the structure of the conversation rather than its volume. We cover the mechanics of that in building feedback loops that actually close.
6. The manager is the system
Every number above converges on one role. Reviews are built mostly from manager observation, delivered by managers, and rated for fairness by the people those managers manage. The capability data is therefore the binding constraint.
- 20% of organisations say their managers are effective at providing coaching and feedback (WTW, September 2025).
- 26% of organisations report managers are very or extremely effective at enabling performance (Deloitte, 2025).
- 54% of managers say they are confident coaching for career development — meaning nearly half are not (Betterworks, via Deloitte 2025).
- 61% of managers could not say they trust the performance management process they administer (Deloitte, 2025).
- 30% of employees strongly agree their manager involves them in goal setting; those who do are substantially more likely to be engaged (Gallup, performance management series).
That last pairing is the cheapest lever in the entire dataset. Goal involvement is a scheduling decision, not a budget line, and only three employees in ten currently experience it.
7. AI arrived in the review cycle faster than the policy did
The single biggest change since the 2024 review season is not a redesign of the form. It is who — or what — is drafting it.
In a survey of 1,342 US managers fielded from 24 June 2025 (ResumeBuilder, via Pollfish; respondents screened for age 25+, household income $75,000+, an associate degree or higher, direct reports, and a company of 11 or more people), 65% said they use AI at work. Of those, 94% use it for decisions about their direct reports.
What AI-using managers use it for
Share of AI-using managers reporting each use. Source: ResumeBuilder/Pollfish, n = 1,342 US managers, fielded from 24 June 2025.
The governance has not kept pace. In the same survey, 32% of managers had received formal training on using AI for people decisions, 43% had informal guidance only, and 24% had none at all.
On the organisational side, WTW found 37% of participants already using AI in performance management, with a similar number considering it — 44% for goal setting, 40% for development plans, 37% within performance reviews themselves and 35% for continuous feedback. Betterworks' 2026 State of Performance Enablement report (n = 2,387; 36% individual contributors, 30% managers, 16% director or above, 18% HR manager or above) adds the perception gap: executives are 6× more likely than employees to believe their review process reflects how AI has changed the work, and only 16% of managers and employees say they understand their organisation's AI strategy at all.
8. What actually moves the numbers
Four interventions in this dataset have a published effect size. None of them is a form.
Meaningful feedback, weekly
80% of employees who received meaningful feedback in the past week are fully engaged (Gallup, n = 13,490 US full-time employees, 10–24 March 2021, ±3pp). The qualifier is "meaningful", not "frequent" — section 5 shows volume alone achieves nothing.
Daily rather than annual cadence
Employees are 3.6× more likely to strongly agree they are motivated to do outstanding work when their manager gives feedback daily rather than annually (Gallup). The realistic intermediate step is quarterly, which carries the fairness effect in section 3.
Involving people in their own goals
Only 30% of employees strongly agree their manager involves them in goal setting, and those who do are markedly more likely to be engaged (Gallup). It costs nothing and is the least-adopted practice in the dataset.
High-quality recognition
Employees who received high-quality recognition were 45% less likely to have left their job between 2022 and 2024, in a Gallup–Workhuman study that followed nearly 3,500 employees longitudinally — a stronger design than a one-shot opinion survey.
The recognition finding is the one most often mis-filed. Recognition is not a softer substitute for performance management; in this dataset it is the intervention with the cleanest longitudinal evidence attached to a business outcome. The full numbers are in our recognition statistics dataset, and the turnover economics in employee turnover statistics 2026.
9. Benchmark cheat sheet
If you need one table to benchmark a performance programme against, this is it. Every figure is the best-sourced value available in 2026 for that question.
| Metric | Benchmark | Source and vintage |
|---|---|---|
| Employees calling the review process fair and transparent | 22% | Gallup panel, Aug 2023, n=18,665 |
| Employees who know what is expected of them | 47% | Gallup panel, Aug 2023 |
| Employees reviewing goals once a year or less | 56% | Gallup panel, Aug 2023 |
| Fairness uplift from quarterly progress checks | 2.1× (22% → ~46%) | Gallup panel, Aug 2023 |
| Engagement uplift from quarterly progress checks | +90% more likely engaged | Gallup panel, Aug 2023 |
| Workers who cannot say they trust the process | 72% | Deloitte GHCT 2025, ~10,000 leaders, 93 countries |
| Managers who cannot say they trust the process | 61% | Deloitte GHCT 2025 |
| Organisations calling their process effective | 39% | WTW, Sept 2025, 280 participants |
| Organisations calling managers effective at coaching | 20% | WTW, Sept 2025 |
| Fortune 500 CHROs whose system inspires improvement | 2% | Gallup, Jun–Jul 2023, n=135 |
| Evaluations based on manager observation | 67% | Gallup panel, Aug 2023 |
| Organisations still using a five-point rating scale | 45% | WTW, Sept 2025 |
| Extra personality-focused feedback received by women | +22% | Textio, 2022 corpus (~25,000 reviews) |
| Attrition attributable to low-quality feedback | ~10% | Textio, 2024 (23,000 reviews, 450 surveyed) |
| US managers using AI in performance assessment | ~59 in 100 (derived) | ResumeBuilder/Pollfish, Jun 2025, n=1,342 |
| Employees fully engaged after meaningful weekly feedback | 80% | Gallup, Mar 2021, n=13,490 |
| Retention effect of high-quality recognition | 45% less likely to have left | Gallup–Workhuman, ~3,500 followed 2022–2024 |
10. Six performance review statistics that should not be cited as published
This topic has a citation-hygiene problem. The following figures are all real in the sense that someone published them, but each is routinely presented in 2026 in a way its source does not support.
| Circulating claim | What it actually is |
|---|---|
| "Managers spend 210 hours a year on performance management" | A CEB (now Gartner) figure quoted by SHRM on 19 August 2015. It predates continuous-feedback tooling, remote work and AI drafting entirely. Usable as a historic order-of-magnitude anchor — which is how it is used in section 3 — not as a current measurement. |
| "58% of executives say performance management drives neither engagement nor high performance" | Deloitte, published in Harvard Business Review in April 2015, alongside the finding that Deloitte itself spent close to 2 million hours a year on the process. Eleven years old and still quoted without a date. |
| "Only 14% of employees are inspired by their reviews / 29% call them fair" | Gallup, from its performance management research series published around 2017–2019 and still hosted on gallup.com without a fieldwork date. Gallup's own fresher measurement — 22% fair and transparent, August 2023, n=18,665 — is the one to cite for 2026. |
| "71% of companies still conduct annual performance reviews, according to SHRM" | Widely repeated across aggregator listicles with no report name, sample size or date attached, and not traceable to a SHRM publication. The defensible employee-side equivalent is Gallup's 56% who review goals once a year or less (August 2023). |
| "Half of managers use AI to decide who gets promoted and fired" | Real survey (ResumeBuilder/Pollfish, n=1,342, June 2025), but the headline compresses two conditional steps: 77% of the 65% who use AI at all. The unconditional figure is about 50%, the respondents were screened for income and education, and it is US-only. |
| "Only 2% of companies say performance management delivers exceptional value" | Mercer's Global Performance Management Study, 2019. Still accurate as a historical statement, and strikingly close to Gallup's 2023 CHRO figure — but the two are different populations answering different questions and should not be presented as a trend line. |
Methodology and sources
Every figure on this page is attributed to a named organisation with its reference period and, where the publisher disclosed it, its sample size. Where a statistic is derived, the inputs and the arithmetic are shown in the body rather than presented as a survey finding. Percentages are reproduced as published and not re-based.
Citation. If you use these figures, please cite the underlying source rather than this page, and link here as the compilation. The three derived metrics — the 1-in-9 two-sided trust gap, the price of fairness in manager-hours, and the 59-in-100 AI-assessed majority — are original to this page and should be attributed to Rewordin together with their stated inputs and assumptions.
The review is annual. Recognition does not have to be.
The clearest finding in this dataset is that frequency and meaning beat process design: weekly meaningful feedback puts 80% of employees in the fully-engaged group, and high-quality recognition made employees 45% less likely to leave over two years. Rewordin delivers gift-card rewards in 150+ countries in local currencies, with the reporting to show who was recognised, when and for what — so the record of someone's year is not written for the first time in December.
About the authors
Maciej is the founder and CEO of Rewordin, a global employee rewards and recognition platform delivering gift cards in 150+ countries. He works directly with HR, People Ops and finance teams on reward programme design, HR systems integration and bulk gift card procurement, and writes about the research behind effective recognition. Based in Wrocław, Poland. Connect on LinkedIn →
Natalia is the CFO of Rewordin and co-reviewer of every cost, compensation and regulatory claim published on the platform — including the manager-hour costing in section 3 and the two conditional-probability derivations in sections 2 and 7, each of which is shown here with its arithmetic and its assumptions rather than presented as a survey finding. Connect on LinkedIn →
What percentage of employees think performance reviews are fair?
22% of US employees strongly agree their performance review process is fair and transparent, in Gallup's U.S. Workplace Panel Study of 18,665 employees fielded in August 2023. An older and more widely quoted Gallup figure puts the share calling reviews "fair" at 29% and "accurate" at 26%, but that research dates from roughly 2017–2019; the 2023 measurement is the one to cite for 2026. Betterworks data cited by Deloitte puts the share who think reviews are very fair and equitable at fewer than one in three.
Do performance reviews actually improve performance?
On the published evidence, the annual review form does not. Only 14% of employees strongly agree their review inspires them to improve, 20% strongly agree their performance is managed in a way that motivates outstanding work, and 2% of Fortune 500 CHROs strongly agree their own system inspires improvement. What does have a measured effect is frequency and quality of conversation: 80% of employees who received meaningful feedback in the past week are fully engaged, and quarterly progress checks make employees 90% more likely to be engaged and 2.1 times as likely to consider the process fair.
How many companies still do annual performance reviews?
The honest answer is that the employee-side measurement is better than the employer-side one. Gallup found 56% of US employees formally review their goals with a manager once a year or less; an earlier Gallup split put 48% on an annual cycle and 26% on less than annual. The widely circulated claim that "71% of companies still conduct annual reviews, according to SHRM" cannot be traced to a SHRM publication and should not be cited. What is documented is that ratings themselves are being adjusted rather than abolished: 45% of organisations in WTW's September 2025 research still use a five-point scale, and 54% have changed or are considering changing their scale.
Are performance reviews biased?
Measurably, yes — and the evidence is textual rather than attitudinal. Textio's analysis of 23,000 performance reviews in 2024, and roughly 25,000 in its 2022 edition, found women receive 22% more feedback about their personality rather than their work, twice as much unactionable feedback as men, and 38% more problematic feedback among high performers. Black employees receive twice as much unactionable feedback as white or Asian colleagues. 78% of women recall being described as emotional at work against 11% of men. Since 67% of evaluations are based primarily on manager observation, this is a defect in the dominant input, not a peripheral one.
How often should performance reviews happen?
At least quarterly, on the only intervention in the published data with a clean effect size attached. Gallup found employees with quarterly progress checks are 90% more likely to be engaged and 2.1 times as likely to say the process is fair and transparent — taking perceived fairness from a 22% baseline to roughly 46%. Moving from annual to quarterly adds three conversations per employee per year: about 1.5 manager-hours per employee annually, or roughly a 6% increase on what an annual cycle already consumes. The finding is correlational, so treat it as the cheapest available experiment rather than a guarantee.
How many managers use AI to write performance reviews?
In a June 2025 survey of 1,342 US managers by ResumeBuilder via Pollfish, 65% use AI at work and 91% of those use it in performance assessments — about 59 in 100 managers in that population overall. 21% of AI-using managers say they let AI decide without human input often or always, which works out at roughly 14 in 100 managers. Only 32% had received formal training on using AI for people decisions and 24% had none. On the organisational side, WTW found 37% of participants already using AI somewhere in performance management.
Do employees trust the performance management process?
Mostly not, and neither do their managers. Deloitte's 2025 Global Human Capital Trends survey of nearly 10,000 leaders across 93 countries found 72% of workers and 61% of managers could not say they trust their organisation's performance management process. Inverting those gives 28% of workers and 39% of managers who can — which means that if the two are independent, both parties to a review trust the instrument in only about 11% of conversations, or one in nine.
What is the best alternative to the annual performance review?
Not a replacement form, but a change of cadence and of who does the talking. The four interventions with published effect sizes are weekly meaningful feedback (80% of recipients are fully engaged), daily rather than annual manager feedback (3.6 times more likely to be motivated to do outstanding work), involving employees in setting their own goals (only 30% currently experience this), and high-quality recognition, which made employees 45% less likely to have left their job across a Gallup–Workhuman study that followed nearly 3,500 people from 2022 to 2024.