How to Run a Talent Review and Build a Succession Plan

What is a talent review, and why does it matter?

A talent review is a structured discussion, by the leaders who know the work, of people's performance, potential and aspirations. Succession planning uses what the review finds to make sure the roles that matter most can be filled when they fall empty. The CIPD's factsheet on succession planning, updated in December 2025, describes the job as identifying and growing talent to fill leadership and business-critical positions in the future.

Done well, it serves three purposes. Continuity: when a critical role empties, the work carries on. Risk: leaders can see where the organisation is exposed before a resignation forces their hand. Growth from within: people see routes to bigger roles, and more vacancies are filled by people who already know the business.

It is not a pay exercise, a ranking or a promotion panel. Keeping those apart is the single most important design choice, because the moment pay or promotion rides on a placement, managers start arguing for outcomes rather than describing evidence.

How do you identify critical roles?

Start with roles, not people. A critical role is one whose absence would seriously harm the organisation and which would be hard to fill well. Seniority often comes with impact, but not always: a scarce specialist or a frontline supervisor can matter more than a senior general manager whose work others can absorb.

Score each role from 1 to 5 on four criteria, put a fact behind each score, and combine them with weights you apply to every role. A common starting point is impact 35%, scarcity 25%, time to fill 20% and risk of loss 20%, with roles scoring 3.5 or more treated as critical. Score the role, not the person in it, and score risk of loss on facts such as announced plans or a known approach from a competitor, never on guesses about age or family.

What is the difference between performance and potential?

Performance is how well someone does their current role this year. It looks back and should already be documented in your review cycle. Potential is evidence that someone could grow into broader or more complex roles in the next few years. It looks forward and is a judgement, which makes it the place where bias does most damage.

Make potential fair by describing it as behaviour you can observe. Useful indicators include learning agility (Michael Lombardo and Robert Eichinger linked high potential to the ability to learn from experience in Human Resource Management in 2000), breadth of judgement, aspiration in the person's own words, drive and resilience, working through others, and curiosity. Claudio Fernández-Aráoz, writing in Harvard Business Review in June 2014, described motivation, curiosity, insight, engagement and determination as markers of potential.

Require one real example for every potential rating: the situation, what the person did and what happened. Confidence in meetings, visibility to senior leaders and similarity to current leaders are not potential. Neither are part-time work, leave, caring responsibilities or age.

How should you use the 9-box grid?

The 9-box crosses performance (columns, from below to above expectations) with potential (rows, from limited to strong evidence). Box 9, top right, is above-expectation performance with strong evidence of potential. A simple numbering rule keeps everyone aligned: box = (potential minus 1) x 3 + performance.

Use it to organise a discussion, not to label people. Three boxes are often misread. Box 3 holds high performers with limited evidence of broader potential, often deep experts who can still be ready-now successors for specialist roles. Box 7 holds people with strong potential signs but low results, who usually need a conversation about what is in the way. Box 5, the core, is where many talent pools should start.

• Never a ranking or a quota per box.
• Never a pay input.
• Never a label shared with the person.
• Never a promise of promotion.
• Never decided by a tool or AI model.
• Never fixed: revisit at least once a year.

How do you run a fair talent review meeting?

The meeting is only as good as the pre-work. Ask every manager for performance evidence, two examples of potential, the person's aspiration in their own words (from a short conversation), mobility, risk of loss with facts, and development ideas, at least five working days before the meeting.

In the room, a chair makes final calls and speaks last; a facilitator runs the clock, asks 'what did you see them do?' and applies bias interrupters, but places nobody. Read the rules of the room aloud: evidence before opinion, performance and potential kept apart, no personal circumstances or protected characteristics as reasons, and nothing shared until HR confirms what each person will be told.

Check placements by group before they are final. Economists Alan Benson, Danielle Li and Kelly Shue studied management-track employees at a large retail chain and found women received lower potential ratings than men despite higher performance ratings, with the difference explaining 30% to 50% of the gender promotion gap. If proposed ratings show a gap, do not fix it with arithmetic: apply the same evidence test to everyone in both groups, and never report a group of fewer than five people.

How do you build succession slates and measure bench strength?

For each critical role, name the people who could fill it and assess each against the role's requirements in one of three bands: ready now, ready in 1-2 years, or ready in 3+ years. Readiness is about the role, not the 9-box. Separately, name emergency cover: who keeps the role running tomorrow if the holder leaves on Friday.

Rate succession risk with a simple rule: low if at least one successor is ready now, medium if the best is 1-2 years away, high if nobody is ready within two years. Track ready-now coverage (critical roles with a ready-now successor divided by all critical roles) and bench strength (named successors per critical role). Use talent pools as well as named slates where several roles share skills; the CIPD factsheet describes both approaches.

Every high-risk role needs a decision this month: name emergency cover, develop someone from a pool, document critical knowledge, redesign the role so it depends less on one person, or hire.

How do you make successors actually ready?

A slate without development is a list of hopes. Give every successor a plan built from their two or three biggest gaps, with at least one real experience such as an acting role, a stretch project or a secondment. The 70-20-10 rule of thumb, which grew out of work by Morgan McCall, Michael Lombardo and Robert Eichinger at the Center for Creative Leadership in the 1980s, is a reminder that most development happens on the job; treat it as a guide to the mix, not a formula.

Give every action an owner, a due date and a measure, fit the plan to the person's working pattern and life, and book two review points now. Then keep the routes open for everyone: publish career paths (dual ladders, lattices, skills-based and frontline-to-supervisor paths), post vacancies internally and give every internal candidate feedback. Track the internal fill rate. If you want to test your own judgement with people data before you start, the People & Performance Data Judgment assessment from AssessAll is one option, and Jobulary can help you turn what you learn into your own development plan.

What should you tell people after a talent review?

Everyone in scope should hear from their manager, in person or on video, within a few weeks of the meeting. Share what matters: how the review saw their performance this year, the strengths people mentioned, the development agreed and what would change the picture. Do not share box labels, other people's placements or anything that sounds like a promise.

Named successors need two messages at once: the organisation is investing in them, and being named is not an appointment or a date. People who hoped to be named and were not deserve the specific reason and a plan to build the missing evidence. When someone asks 'what box am I in?', explain that the grid is a snapshot used inside the meeting and move straight to the evidence and the development, which is what they really want to know.

Keep a short record of what each person was told. It protects the person, the manager and the process if anyone later asks what was said.

Where does AI help, and where must it never decide?

AI tools can summarise anonymised evidence, sort a manager's notes under potential indicators, draft development plans and messages, flag biased language and spot inconsistent placements. Use approved tools and anonymised or approved data only.

AI must never rate potential, judge readiness, predict who will leave or select anyone. Rules are tightening: under the EU AI Act, AI systems used to make decisions on promotion or to evaluate the performance and behaviour of workers are classed as high-risk in Annex III. Check the current dates and duties where you operate. A person decides every talent outcome and can explain it in plain words.

The Talent Review and Succession Planning Kit from Bodhih Training includes a workbook that calculates every measure in this guide, the forms and meeting guide, and dated country notes on privacy and AI.