An assessment centre can give you rich, job-relevant evidence about how someone will perform, but only if it is built properly. Good assessment centre design starts with the job, measures each competency more than once, uses trained assessors and combines the evidence in a disciplined way. A poorly designed centre looks impressive on the day and tells you very little. This guide is for HR leaders, talent teams and line managers in the GCC who are commissioning a centre or reviewing one they already run. It draws on the international guidelines for assessment centre practice and the British Psychological Society (BPS) standard. What an assessment centre is The International Taskforce on Assessment Center Guidelines (2015) lists ten essential elements. In summary, a true assessment centre has: competencies identified through job analysis, with behaviour classified into them; multiple assessment components, including job-related simulations, linked to competencies in a matrix; multiple trained assessors per participant, who record behaviour systematically as it happens; a defined method for integrating ratings, by discussion or statistically; standardised conditions, so every participant has the same chance to perform. A single interview plus a personality questionnaire is not an assessment centre. The same model is used for development centres, where the aim is feedback and growth rather than a decision. When an assessment centre is worth it Centres take time and money, so use them where the decision justifies it: senior or hard-to-reverse appointments, promotion into a first leadership role, graduate and national talent programmes where candidates have little work history, and development programmes that need behavioural feedback. On predictive value, a re-analysis of selection research by Sackett and colleagues (2022) estimated a mean operational validity of .29 for assessment centres. A follow-up paper (Sackett et al., 2023) noted this rises to .33 under a different assumption about criterion reliability, and the same work ranked structured interviews highest, at .42. The lesson for buyers: a centre is not automatically better than a good structured interview. Its value depends on design quality and on the behavioural evidence it gives for development. The assessment centre design process 1. Job analysis Study the role through interviews with job holders and managers, critical incidents, documents and future priorities. The output is a short list of behaviours that separate effective from less effective performance. The BPS standard (2015) says the number of criteria must be no greater than can be assessed effectively, noting research that points to a maximum of between 6 and 12. 2. Competency framework Define each competency with observable positive and negative indicators. If you already have a framework, select the competencies that matter for this role rather than assessing all of them. 3. Assessment matrix The matrix shows which exercise measures which competency. The BPS standard requires each criterion to be assessed by at least two different methods, at least one of them an exercise, and ideally by two different exercises. It also suggests limiting each exercise to around four criteria, to reduce the load on assessors. Sample assessment matrix (middle-manager role) Competency Group exercise Role play Case / in-tray Presentation Structured interview Psychometrics Strategic thinking X X X Decision making X X X Influencing and communication X X X Leading people X X X Stakeholder management X X X Delivering results X X X Competencies per exercise 4 3 3 3 3 2 Each competency appears at least twice and at least once in a simulation, and no exercise carries more than four. 4. Exercise types Group exercise: a shared problem that shows influencing, collaboration and decision making. Role play: a one-to-one meeting, such as a performance conversation, with a trained role player following a consistent brief. In-tray or case analysis: realistic emails, reports and data to prioritise and act on. Strong for judgement and planning. Presentation: a prepared recommendation followed by questions. Structured interview: the same competency-based questions for everyone, scored against anchors. Psychometrics: ability tests and personality questionnaires with documented reliability, validity and suitable norms, interpreted by a qualified user. Make simulations reflect the real job and the regional context: realistic stakeholders and the kinds of decisions the role takes. 5. Rating scales and behavioural anchors The 2015 guidelines describe behavioural observation scales, checklists and behaviourally anchored rating scales (BARS). A BARS gives examples of effective, average and ineffective behaviour at each scale point, so a "4" means the same to every assessor. Write anchors from the exercise content and pilot them. Assessor selection and training The guidelines require multiple assessors per participant and encourage a diverse assessor group. The BPS standard sets a minimum of one assessor for every three participants, ideally one for every two. Senior line managers often make credible assessors if they are trained and can commit the time. Training should cover the competencies, the exercises, the observe, record, classify and evaluate process, common rating errors and confidentiality. For inexperienced assessors, the guidelines suggest around two days of training per day of exercises, while noting that length and quality are not the same. Frame-of-reference training Frame-of-reference (FOR) training gives assessors a shared view of what each competency means and what each performance level looks like. Assessors rate sample performances, compare with expert ratings and discuss differences until they calibrate. A meta-analysis by Roch and colleagues (2012) found FOR training improves rating accuracy across settings. Include it in every assessor briefing, including refreshers. Wash-up and data integration The wash-up combines evidence into competency and overall ratings. The guidelines allow consensus discussion or statistical aggregation, provided the result is reliable. The BPS standard is more specific for selection: arithmetic approaches shall be used for the overall centre rating, and any final decision that departs from it must be explained. Discussion remains useful for developmental feedback. Good practice: assessors finalise exercise ratings independently first, the chair ensures every assessor's evidence is heard, discussion stays on recorded behaviour, and decisions are documented. Candidate experience and feedback Tell participants in advance what the centre involves and how their data will be used. The BPS standard requires that all participants are told of decisions based on their performance, and recommends feedback ideally within two weeks and no later than four. For internal and national candidates, good feedback protects engagement even when the answer is no. Fairness and adverse impact The BPS standard defines adverse impact as a significant difference in the success rates of different groups, requires that exercises do not unfairly help or disadvantage any subgroup, and asks for demographic data for monitoring. A meta-analysis by Dean, Bobko and Roth (2008) found ethnic subgroup differences in US samples, while women on average scored slightly higher than men. Findings from elsewhere will not transfer directly to the GCC, so monitor your own data by gender, nationality and language group, and plan reasonable adjustments for disabled participants. Employment and data protection rules differ across GCC states, so check obligations with a qualified adviser. Practical GCC considerations Bilingual delivery (Arabic and English) Decide early whether participants choose their language and whether assessors can rate in both. The International Test Commission guidelines on translating and adapting tests (second edition) stress checking that an adapted version measures the same thing. The 2015 assessment centre guidelines mention translation and back-translation, and say observation, classification and systematic integration must stay standardised after adaptation. Pilot both versions, and do not mark candidates down for fluency unless it is a genuine job requirement. Nationalisation programmes Many GCC employers use centres within national talent programmes. In the UAE, the Ministry of Human Resources and Emiratisation requires private companies with 50 or more workers to raise the share of Emiratis in skilled roles by 1% every six months. Saudi Arabia's Nitaqat programme classifies firms into bands by their Saudisation rate. Centres for national graduates and future leaders should lean towards clear development feedback and links to individual development plans. Rules change, so confirm current requirements with the relevant ministry or an adviser. Virtual centres Virtual delivery suits multi-country programmes. The BPS Guide to Virtual Assessment Centres (2024) supplements its standard, covering technology, validity, design, assessors and participants. The 2015 guidelines warn that online delivery can threaten standardisation. Test the platform, run a practice session, plan for connection failures and check that virtual exercises still produce observable behaviour. Common mistakes Assessing too many competencies, so each is measured poorly. Generic exercises that do not reflect the role or region. Cutting assessor training, especially calibration. Letting the most senior person in the wash-up decide. Combining selection scores by "gut feel". Late, vague or missing feedback. Never checking whether ratings predict performance or show group differences. Assessment centre design checklist Use this before signing off a centre, or ask a supplier for evidence on each point. The BPME toolkit has further resources. Purpose agreed in writing: selection, promotion or development. Job analysis documented for this role. Competencies defined with observable indicators, kept to a manageable number. Each competency in at least two methods, at least one a simulation; about four competencies per exercise at most. Exercises realistic for the role and region, and piloted. Behaviourally anchored rating scales tested. At least one assessor per three participants, ideally per two. Assessors trained, including frame-of-reference calibration, and signed off. Integration method set in advance, arithmetic for selection. Arabic and English versions adapted and piloted, if both are used. Virtual platform tested, with a backup plan. Feedback planned within two to four weeks. Demographic monitoring and a plan to review adverse impact and validity. Find the right expert on BPME Designing a centre well takes specialist skill in job analysis, exercise writing and assessor training. On BPME you can find a business psychologist or assessment specialist , compare profiles, services and availability, and message professionals directly, free. You can also browse assessment services in the marketplace or read more in Insights .