You want recognition to reward the right things. Which criteria can you actually measure, and does recognition tied to them work?

Most of what is known about appreciation at work comes from asking people whether they feel appreciated. A 2025 systematic review of employee appreciation brought together 41 published records. Thirty-nine of them reported 43 empirical studies, 10 qualitative, 27 quantitative and 6 mixed-method; the other two were an earlier systematic review and a theoretical essay. Of the 27 quantitative studies, 19 were one-off surveys, six followed people over time, one was a diary study and one was an experiment. Twenty-four of the studies that used quantitative methods measured appreciation through employees' own questionnaire answers, and the instrument used most often is, in the authors' words, "unpublished and to date unvalidated". Supervisors were the most discussed source of appreciation (30 records), ahead of co-workers (12), top management (11) and customers (4). Feeling appreciated went with more positive feelings at work, better mental health, and more engagement and commitment, although one study found no relation with perceived stress. None of the instruments measured appreciation for a person's individual characteristics; they measured appreciation in general, or for positive work behaviour. The authors call for more experiments, and list how best to communicate appreciation as a question for future research. The studies came from 14 countries, most of them from the United States, Germany and Switzerland; none was from the Gulf. (Read in library)

A programme that tied recognition to an audited standard, set in advance, improved safety climate slightly in a randomised trial. On eight commercial construction sites in the Boston area, matched in pairs, one site of each pair was randomly assigned to a programme with three parts: a weekly inspection by a trained safety manager, which scored each site and each subcontractor by the percentage of safe observations; weekly posters and reports showing the scores; and monthly recognition whenever the site's score beat its goal. The goal, between 94.8% and 96.3% depending on the site, was the median monthly score over the previous twelve months at sites of similar size and scope, set, in the authors' words, to be "fair, consistent, attainable, and competitive". Beating it earned the whole site a catered lunch and a raffle for a parking pass or a fuel-station gift card. Among the 615 workers surveyed both before and after, safety climate rose 0.5 points on a 50-point scale at programme sites and fell 0.8 points at the others; after adjustment the difference was 1.64 points (p = 0.01), and workers at programme sites described more safety awareness, communication and teamwork. The authors chose these "leading" indicators on purpose: schemes that reward injury-free records, they note, may discriminate against injured workers and reduce injury reporting. The limits are real. The researchers developed the programme with local industry partners and evaluated it themselves. Recognition was tested together with inspection and feedback, so its own contribution cannot be separated out. The trial was not large enough to detect a change in injuries. The site populations differed despite randomisation, researchers were on all eight sites almost daily, and the findings apply to sites that already ran safety programmes. (Read in library)

The largest study links recognition to engagement, but its measure of recognition includes pay. In the 6th Korean Working Conditions Survey, carried out in 2020–2021, 25,285 employees answered questions about their workplace. Of the workplace factors tested, recognition had by far the strongest association with engagement: a path coefficient of 0.394, against 0.123 for fairness, 0.070 for involvement and 0.039 for transformational leadership. Recognition here was three statements: that people receive appropriate recognition for their work; that employees receive recognition and praise when they do their job well; and that, considering their effort and achievements, they receive appropriate compensation. So the strongest association in the study belongs to a mix of praise and pay. Everything was measured once, and the authors state that the cross-sectional design restricts causal inference. (Read in library)

A programme can measure the right things and still leave its aims unclear to staff. An evaluation of a staff reward and recognition programme at a large fashion retailer in Cape Town, published in 2012, interviewed its organiser and ten participants. On paper the criteria were measurable. Each month, stores earned points for reaching a percentage of their sales target, for service ratings gathered by contacting randomly selected customers, and for new accounts opened compared with the previous year; the points were shared equally among a store's staff, and managers could add points for individuals they saw excelling. Yet eight of the ten employees said the programme's outcomes were not well defined, and one said they had never been told how it worked or what its incentives were. Delivery varied between sites, staff questioned how the programme's activities related to performance, and all ten criticised the reward, a three-day trip to Cape Town for people who lived there. The organiser and seven of the ten participants agreed that better customer service was the programme's aim; the other three were unsure, naming aims that ranged from self-development to less pilfering. The evaluators recommended redefining its outcomes as standard performance indicators and monitoring them. This is eleven interviews in one company, and a confidentiality clause limited the evaluators' access to programme records. (Read in library)

The figures most often quoted on recognition come from companies that sell it. Gallup, which sells engagement surveys and workplace consulting, reported with Workhuman, which sells employee recognition software, that employees it classed as well recognised were 45% less likely to have left their organisation two years later, from tracking nearly 3,500 employees from 2022 to 2024. The article does not say how turnover was measured or whether anything else that differed between the groups was taken into account, so it shows that people classed as well recognised stayed more often, not that recognition kept them. (Gallup article) Achievers, which sells recognition and rewards software, publishes a guide to five measurable criteria for employee recognition. Its five are performance, everyday behaviours such as helping a teammate or staying calm under pressure, company values, milestones, and tracking whether recognition reaches people. Its two statistics come from a recognition report on Achievers' own website, one of them that 91% of employees say they would put in more effort if they felt valued, and it cites no independent study. (In our library) In both cases the publisher sells products or services in this field, so neither source is independent evidence.

Does recognition lift engagement, or do engaged people notice more recognition? Every association above could run either way. The Korean survey, and almost every study in the review, measured recognition and its supposed effects at the same moment and from the same people. Gallup's figure follows people for two years, but the article describes no adjustment for anything else that differed between the groups it compares. The review found one experiment among 43 studies. Among the four studies we rely on here, the causal evidence is one trial, and it tested recognition together with inspection and feedback.

Measurable is not the same as good. The construction trial chose inspection scores, meaning hazards controlled and safe practices observed, because the obvious measurable criterion, an injury-free record, can reward not reporting injuries. The retail programme measured sales, customer ratings and new accounts, and still eight of the ten staff interviewed said its outcomes were not well defined. A criterion has to be checkable, but it also has to be something people control, something they know about, and something they cannot meet by hiding a problem.

Outside our library, the experiments pull in opposite directions. In the field studies listed under Further reading, a thank-you card and a purely symbolic award raised output in short, controlled jobs done mostly by students, while an attendance award at an industrial laundry was gamed and cost productivity among the best workers. They are outside our library, so they are named and linked rather than counted.

The advice managers hear most is the least tested. Guidance to make praise specific, timely and sincere is everywhere. The review treats how best to communicate appreciation as a question for future research, not a finding.

Recognise people against criteria someone else could check, and treat the claim that recognition by itself drives engagement or retention as unproven. In the one trial among our four sources, a site was recognised only when an audited score beat a target fixed in advance, and safety climate moved a little. The rest of the evidence says that feeling recognised goes with engagement and wellbeing. That is worth knowing, but it does not tell you what causes what.

Three qualifications keep this honest. The evidence is mostly self-report, gathered once: 19 of the 27 quantitative studies in the review were one-off surveys, and the largest study's measure of recognition includes pay. Measurable criteria can still backfire: choose ones people control, that they know, and that cannot be met by hiding a problem. The most quoted numbers come from sellers: treat a figure from a company that sells recognition software as marketing until you have seen its method and its comparison group.

⚠️ None of these studies was conducted in the Gulf. The review's 14 countries include none from the region, the largest study is Korean, the trial American and the programme evaluation South African. The direction is worth acting on; the sizes are not local.

1. Write down what you recognise people for, and test each line. Could someone else check it and reach the same answer? Replace "above and beyond" with a count, a rate against a target, an observation against a checklist, or a rating from customers you did not choose.
2. Set the bar before you start, from your own history. The construction trial fixed its goal at the median of the previous twelve months' scores at similar sites, so it could be reached but was not automatic.
3. Tell people the criterion and how it is scored, then check they can say it back. Ask five people what earns recognition. If their answers differ, you have a slogan, not a criterion. In the retail programme, eight of the ten staff interviewed said its outcomes were not well defined.
4. Do not reward the absence of bad news. An injury-free, complaint-free or error-free record can be met by not reporting. Recognise what people do to prevent problems, such as practices observed or checks completed, as the construction trial did.
5. Keep a simple record of who was recognised, for what, by whom and when, and look at who never appears in it.
6. Decide how you will know it worked before you begin. Write down today's value of the criterion and the date. After three months, compare it with the same measure in a team that did not change, if you have one, and with its own history if you do not. Then check the work the criterion does not cover: if the recognised number rose while something else fell, people have learned the scheme, not the job.

We found no Gulf study of recognition that we could rely on. The review's 14 countries include none from the region, and none of the other studies in this brief was conducted here.

Checkable criteria are also harder to apply unevenly. Offered as reasoning, not as a finding: where nationals and expatriates work side by side on different contracts, a criterion everyone can check leaves less room for favour than a manager's impression. Our brief on appraisal fairness suggests pulling rating distributions by gender and by nationality. Do the same with your recognition record. If nobody has ever looked, that is the finding.

Whether praise from the top, from peers or from a points system lands the same way here is unknown. The review names cultural differences, including how hierarchical a workplace is, as a question for future research. That is a reason to test your own scheme, not to skip the criteria.

We hold no Gulf study on employee recognition. That is an evidence gap, not a settled matter.

The four research sources below are open-access, licence verified per article. None was published or funded by a company that sells recognition software.

1. Employee appreciation: a systematic review and research recommendations — Annals of the International Communication Association
2. Improving safety climate through a communication and recognition program for construction: a mixed methods study — Scandinavian Journal of Work, Environment & Health
3. The impact of recognition, fairness, and leadership on employee outcomes: A large-scale multi-group analysis — PLOS ONE
4. A formative evaluation of a staff reward and recognition programme — SA Journal of Human Resource Management

Cited for what they claim, published by companies that sell recognition or workplace services:

• Employee Retention Depends on Getting Recognition Right — Gallup, with Workhuman, 2024
• 5 measurable criteria for employee recognition to shape behavior — Achievers, 2026

Further reading (named and linked, not republished):

• Employee Recognition and Performance: A Field Experiment — Management Science, 2016. People hired for a three-hour data-entry job, mostly students, were given an unannounced thank-you card after two hours. Output rose afterwards, most when only the top three of eight were thanked, and mainly among those who were not (working-paper version: 7.3% overall, more than 10% among those not thanked).
• Getting More Work for Nothing? Symbolic Awards and Worker Performance — American Economic Journal: Microeconomics, 2011. Students hired by an international non-governmental organisation for a two-hour database task were about 12% more productive when a purely symbolic congratulatory card was announced for the best performers, with no loss of quality (working-paper version).
• Motivational Spillovers from Awards: Crowding Out in a Multitasking Environment — Organization Science, 2016. An attendance award at one of five industrial laundry plants, a monthly draw for a gift card among workers with no unexcused absence or lateness, reduced lateness but was gamed: unplanned single-day absences rose by half, the most productive and most punctual workers lost 6–8% of their productivity, and plant productivity fell 1.4% (working-paper version).
• A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation — Psychological Bulletin, 1999. Across 128 laboratory studies, mostly with children and college students, positive feedback increased later interest in a task, while expected tangible rewards reduced it.