This page lays out a complete performance management framework for CPOs and People leaders: how to define performance, score it consistently, set a talent bar by seniority, run a quarterly review and calibration cycle, and connect grades to promotion, development and pay. It is adapted from the Performance Management Playbook, the method behind Revolut’s performance culture. The detailed playbooks come from QuantumLight; Ash did not author them.
Want it as a PDF? The full Performance Management Playbook is free with your name and work email. Everything below is readable without a form.
The three components of performance
Performance is broken into three parts that are measured separately, because each one answers a different question:
- Deliverables: can this person complete tasks? Scored on speed (meeting deadlines), quality (number of iterations needed) and complexity (how much precedent exists for the task).
- Skills: the technical competencies a role requires. The relevant leader defines them per role, for example the CTO defines skills for engineering roles.
- Culture: how well someone displays company values. Each value is linked to observable behaviours so culture does not stay abstract.
Someone with the right skills who damages team dynamics is not “fine”, and neither is a great culture fit who never ships. You need all three lenses, and skills versus the next level’s requirements tell you whether someone is ready for promotion. Results are quantified into a numeric score and then a grade on a five-level scale: Poor, Basic, Intermediate, Advanced, Exceptional.
Example: deliverables criteria on the five-level scale
| Level | Speed (deadlines) | Quality (iterations) | Complexity (precedent) |
|---|---|---|---|
| Poor | Usually delivers late and misses important deadlines | Output quality is unacceptably low | Only routine tasks done many times before |
| Basic | Usually on time, rarely misses deadlines | Acceptable standard, sometimes needs several iterations | Mix of standard and non-standard tasks |
| Intermediate | Always meets deadlines, often beats them | High standard, typically one iteration | Many non-standard tasks with few comparisons |
| Advanced | Very fast, habitually beats deadlines | Always one iteration, exceptionally high standard | Highly complex tasks with little precedent |
| Exceptional | Identifies important targets, beats deadlines by huge margins | Links any task to the business vision and designs the process to execute it | Exceptionally difficult work that needs radically new processes |
Scorecards: remove bias, give clarity
Consistency is what makes a performance system credible. If managers apply different standards, employees cannot tell what they need to do to progress. The fix is to standardise every assessment with scorecards: skills, deliverables and culture are assessed by managers through a series of true or false statements about observable behaviour. Because the questions are the same across the company, manager bias is limited and employees see exactly what to improve.
The talent bar by seniority
The talent bar is the minimum standard for deliverables, skills and culture at each level. Junior is not senior: compare people with the bar for their level, not with how pleasant they are. Without a published bar, Manager A’s “4” and Manager B’s “4” mean different things and calibration turns into politics. Promotion increases seniority while keeping existing titles, and consistent A-player performance, not tenure, triggers eligibility.
| Level | What the bar looks like |
|---|---|
| Junior | Delivers tasks with a little guidance; learns from mistakes; proactively looks for new tasks. |
| Mid | Independent contributor; delivers team-level KPIs; works on loosely defined problems and owns processes and metrics. |
| Senior | Solves problems independently with little guidance; supports and pushes others; role model for company values. |
| Lead | Owns company-level KPIs and manages a team; performs well above the talent bar; improves how the team hires. |
| Director | Manager of managers; designs processes from scratch to deliver a specific objective. |
| VP | Solves unique company-wide problems; single point of authority for an area across departments. |
See also: Publish a talent bar by seniority.
Sorting the room into three groups
| Group | Playbook guideline weight | Suggested action |
|---|---|---|
| A-players: deliver well above the standard for their role | 15-25% | Promote and compensate strongly |
| Above bar: contribute steadily, room to grow | 65-85% | Develop; push some to become A-players |
| Underperformers: fail to meet expected standards | 0-10% | Exit or performance improvement plan |
These weights are the playbook’s guideline. Read the evidence and caveats before copying them.
Quarterly review and calibration process
Reviews run quarterly. The process has three steps:
- Performance reviews (about 1 to 2 weeks). The performance team sends review requests to line managers. Managers gather input from other managers and fill in the scorecards for their teams.
- Grade calculation and calibration (about 1 week at most, including leadership review). Grades are calculated automatically from the scorecards, then the performance team calibrates them. Senior leadership reviews and approves the final grades and spots teams that need more attention.
- Announce results and deliver feedback. Employees receive their results and discuss them with their manager in their weekly 1:1. Results can lead to a promotion path, development, a performance improvement plan or a pay change.
What calibration checks
- Generous graders. If half of one team is rated as A-players, the performance team asks the manager to identify the strongest performers so the top band stays credible.
- Missing or thin evidence. If some scorecards are not filled in, or justifications are weak, follow up with the manager straight away so the cycle is not delayed.
- Outliers and sudden jumps. If someone previously graded above bar receives an Exceptional score in one area, examine whether the improvement is genuine or a grading mistake.
Related: how to run a performance calibration meeting, Gartner 2026 on reviews and calibration and why annual reviews fail.
Performance review template outline
Use this structure to build a review template in a spreadsheet or in software. The wording of each statement is yours to write for your roles and values; the five-level criteria above show the style.
| Template section | What it contains | Who fills it in |
|---|---|---|
| Header | Employee, role, seniority level, review period, reviewer | Performance team (pre-filled) |
| Deliverables scorecard | Speed, quality and complexity statements answered yes or no, with a one-line evidence note each | Line manager |
| Skills scorecard | Role-specific skills (defined by the relevant function leader), each with level statements answered yes or no | Line manager, with input from other managers |
| Culture scorecard | One block per company value, with observable behaviours at each level answered yes or no | Line manager |
| Calculated grade | Numeric score converted to the 5-level scale: Poor, Basic, Intermediate, Advanced, Exceptional | Automatic from the scorecards |
| Calibration notes | Adjustments, reason for each, flags for missing evidence or outliers | Performance team |
| Outcome and next step | Promotion eligibility, development focus, or improvement plan; feedback delivered in the next 1:1 | Manager, after leadership approval |
Want a filled-in version? Use the performance review template with worked examples (rating scale, goals, evidence and calibration notes).
Tip: start with two critical job families, write their talent bar and scorecards, run one cycle, then extend.
Moving off a yearly cycle? Read alternatives to annual performance reviews and how to calibrate performance reviews.
Performance management framework for scaling startups: worked examples by company size (illustrative)
Illustrative only. These three setups are suggested starting points built from the playbook method. They are not benchmarks, and nothing here is drawn from a survey of companies. Adjust names, timings and roles to your context.
About 50 employees
| Element | Sample setup |
|---|---|
| Cadence | Quarterly review, kept light: managers fill one scorecard per person, then a single calibration meeting of about 60 to 90 minutes. Promotions decided every 6 months. |
| Roles | Founder or CEO approves grades. Each manager completes scorecards. One person (often the head of people or operations, part-time) runs the calendar and facilitates calibration. |
| Decision points | Do grades feed pay, or only development, in the first two cycles? Who breaks a tie when two managers disagree? Which two job families get a written talent bar first? |
| What to skip | Percentage bands (a handful of people per band makes quotas meaningless), a talent bar for every level, multi-step approvals and a separate performance team. |
About 250 employees
| Element | Sample setup |
|---|---|
| Cadence | Quarterly cycle: about 1 to 2 weeks for scorecards, then calibration by department followed by one cross-department session, within about a week including leadership review. Promotions every 6 months. |
| Roles | Managers complete scorecards. A small performance or HR business partner team runs the cycle and calibrates. Function leaders define skills for their roles. The leadership team approves final grades. |
| Decision points | Which job families have scorecards at launch? Who approves promotions (for example department heads, after a manager business case)? What happens below the bar: improvement plan, separation or both? |
| What to skip | Forced distribution, a bar for every level in every function on day one, and elaborate peer-feedback programmes. Start with the job families that matter most and add others after one cycle. |
1,000+ employees
| Element | Sample setup |
|---|---|
| Cadence | Quarterly, staggered by function so the performance team is not calibrating everything in the same week. Layers: team, then department, then company-level review. |
| Roles | Managers fill scorecards. HR business partners and a central performance team calibrate and check consistency across departments. Function leaders own skills definitions. Senior leadership approves. |
| Decision points | Do you use percentage bands as a quota or only as a check that prompts questions? How do you handle manager-level generosity patterns? How are scorecards kept consistent across very different functions? |
| What to skip | Manual spreadsheets for grade calculation, a single scorecard for every function, and band quotas applied strictly at team level (small teams make the numbers noisy). |
Whatever the size, the one-page review-cycle checklist gives you the steps in order, and the calibration meeting guide covers the running order.
From grades to outcomes
- Promotion. The playbook runs promotion every 6 months. Its example rules: at least 1 year at the current level, no review below the talent bar, and A-player grades in at least 2 of the last 4 reviews (4 of 4 for promotions above Mid). Managers validate with a business case, and department heads approve.
- Underperformance. Offer a choice as soon as someone falls below the bar: an enhanced separation package or a 6-week performance improvement plan with SMART targets, managed by the manager with HR oversight.
Evidence and caveats
The A-player band (about 15-25%) comes from the playbook itself and is a guideline from Revolut’s own experience, not an industry benchmark. We could not find a neutral benchmark for those exact percentages. What published research does show:
- GE’s “vitality curve” is the best-known precedent: top 20%, middle 70%, bottom 10%, with the top group receiving raises two to three times larger and the bottom group often fired. Source: Precarious curve ahead: the effects of forced distribution rating systems on job performance (Human Resource Management Review).
- Hard caps carry a retention risk. At a multinational pharmaceutical company with a cap of 1 in 5 on top ratings, employees nominated for the top tier but not awarded it were at least 34% more likely to leave voluntarily. Source: Bond, Cut to the Curve (Management Science, 2025); summary in the Cornell Chronicle.
Practical reading: treat the bands as a starting point, keep calibration evidence-based, and explain decisions to the people near the cut-off.
When a framework or forced distribution is the wrong approach
A structured framework is not free, and a forced distribution carries more risk than a framework does. Consider holding back, or changing the design, in these cases:
- Very small teams. When a band represents only a few people, a percentage target pushes you to rank people to fit the number. Calibrate against the bar and discuss each case instead.
- Work that is hard to attribute. Where outcomes come from tight team effort, individual scorecards on speed and quality can misrepresent who contributed. Weight team outcomes and use observed behaviours.
- Managers without time or evidence habits. If scorecards turn into box-ticking, the framework creates paperwork and a false sense of precision. Fix the manager routine (regular 1:1s, short evidence notes) before adding process.
- Hard caps on top ratings. Capping the top tier means some people who meet the standard are denied it. One study found a retention cost for those near-misses (see below). If you use bands, treat them as a prompt for questions during calibration, not a quota.
- Unstable roles or very early-stage companies. If job content changes every few months, a detailed talent bar goes out of date quickly. Write a short bar for the few roles that are stable and revisit it often.
- Using ratings as the only input to pay or exits. If grades drive pay or separation, the quality of evidence and the appeal route matter more. Add a human review before any adverse outcome.
Sourcing caveats. The percentage bands on this page are the playbook’s own guideline, not an industry benchmark. The retention figure comes from a single study at one multinational pharmaceutical company (Bond, Management Science, 2025), so it may not transfer to your organisation. The GE vitality curve source is a review of forced-distribution systems, not a recommendation. We did not find neutral data comparing frameworks like this one against having no framework, so the reasoning above is judgement, not measured fact. Ash works for Revolut People, which sells performance software, so weigh this section with that in mind.
Running it in software
Comparing tools? Read the performance management software buyer’s guide for neutral evaluation criteria.
The playbook is the method. Software such as Revolut People is how teams run goals, skills and culture (the talent bar), reviews, calibration and outcomes in one place instead of slides and spreadsheets. See Revolut People performance management for the product.
Guides and templates
- Performance review template with worked examples
- How to run a performance calibration meeting
- Employee performance rating scales explained
- Performance management software buyer’s guide
- Performance review cost calculator
- HRIS switch checker
- Posts: goals that don’t cascade, publish a talent bar by seniority, year-end reviews and the HRIS stack, all blog posts.
FAQ
What is a performance management framework?
It is the written system that defines what good performance is, how it is scored, who reviews it and what happens next. This framework uses three components (deliverables, skills and culture), standardised scorecards, a talent bar for each seniority level, a quarterly review and calibration cycle, and clear links from grades to promotion, development, improvement plans and pay.
How often should performance reviews and calibration run?
The playbook runs reviews quarterly: managers complete scorecards in about 1 to 2 weeks, the performance team calibrates in about a week at most (including senior leadership sign-off), and results are announced and discussed in 1:1s. Promotions are decided less often, every 6 months, so that strong performance has to be sustained beyond a single review.
What is a talent bar?
A talent bar is the minimum standard for deliverables, skills and culture at each seniority level (for example Junior, Mid, Senior, Lead). People are compared with the published bar rather than with each other or with how likeable they are, which keeps one manager's rating comparable with another's.
What happens in a calibration meeting?
The performance team checks the grades produced by the scorecards before they are final. It looks for managers who rate too generously (for example half of one team rated as top performers), missing scorecards or thin evidence, and sudden jumps or unusual highs and lows that need explaining. Senior leadership then reviews and approves the final grades.
Are the A-player percentages a rule I should copy?
No. The 15-25% A-player band, 65-85% above bar and 0-10% underperformer split are the playbook's own guideline, not an industry benchmark. Published research on capped or forced rating curves is mixed, so treat the bands as a starting point, check them against your own context, and make sure grades are backed by evidence.
Can I get the framework as a PDF?
Yes. The full Performance Management Playbook PDF is free with your name and work email via the form linked on this page. Everything in the framework itself is readable here without a form.
Can software run this process?
Yes. Teams can run goals, skills and culture, reviews, calibration and outcomes in a performance platform such as Revolut People instead of slides and spreadsheets. The method works without software, and software does not replace the judgement in calibration.