Start from the outcome you are buying and the people problem you are solving. The right system for an ambitious organization is the wrong one for an organization that needs reliability, and neither is decided by what the form looks like.
For the CHRO, HR leader or business leader who owns the performance system. Article 7 of seven in the performance management investigation.
The leadership team has agreed that performance management needs to change. One person wants OKRs. Another wants fewer reviews. Someone proposes dropping ratings, and someone else wants stronger calibration.
All four are responding to something real. None of them is a problem statement, and a redesign that begins here usually ends with a new form and the same complaints.
Start with the outcome you are buying
Three organizations can want three different things, and all three are legitimate.
Ambition. You want people attempting work they would not otherwise attempt, and you can tolerate missing. That argues for stretch goals, short cycles, and a tolerance for failure that has to be real in pay decisions, not only in the deck.
Reliability. You want the same standard delivered every time, in a business where variation is expensive or dangerous. That argues for clear expectations, tight feedback, and much less enthusiasm for stretch.
Compliance and defensibility. You are in a regulated or litigious setting and need decisions that stand up to scrutiny. That argues for documentation, consistent standards and calibration, and it makes a rating genuinely useful.
As practitioners we have observed a recurring failure pattern: the organization wants the first and builds for the third.
Name the people problem
Then get specific about what is actually wrong, because the mechanisms differ.
| The gap you see | What it usually means | Where to look first |
|---|---|---|
| Productivity | Effort is going to the wrong work, or obstacles are not being cleared | How goals connect across teams |
| Clarity | People cannot say what matters most this quarter | What the label actually changes |
| Engagement | People do not see how their work connects, or feel judged unfairly | What happens after feedback |
| Fairness and trust | Decisions cannot be explained to the people they affect, which has a literature of its own 4 | What the rating is deciding, and what calibration does |
| Capability | The work has changed and the skills have not | Development and hiring, not the appraisal form |
A productivity gap and an engagement gap can look identical in a survey and need opposite responses. The first usually needs someone with authority to clear a path. The second usually needs managers who can explain a decision.
Who does this system have to be able to remove?
This is the question leadership teams avoid in the design meeting and argue about later, and it decides more about the system than any scale.
Organizations sit somewhere on a range, and all of the positions are legitimate.
| Position | What it implies for the system |
|---|---|
| We develop, we rarely exit | Needs a strong improvement path, documented support, and patience measured in quarters. The system must show what was tried |
| We act, but only on clear evidence | Needs consistent standards, a record that survives challenge, and calibration worth the name |
| We do not carry underperformance past a quarter | Needs fast, frequent judgment, an unambiguous bar, and managers willing to say it out loud |
As practitioners we have observed that many organizations declare the first and behave like the second, while some run the third and are clearer to work in than either.
Then the harder question, and the one to answer explicitly: can this system identify the people adding least value, in a way that matches what the organization says it values? If your values emphasize collaboration and your system measures individual output, it will remove the wrong people and keep the ones who optimized for the number.
Two practical tests. Take the last three exits for performance: could the reasons be shown to the person, and to a colleague, without embarrassment? And take the three people everyone quietly knows are struggling: does your system surface them, or does it surface whoever had the weakest manager?
A system that cannot answer either is not managing performance. It is documenting it.
What else decides the right system for you
Culture is the largest variable, and it is not the only one.
Manager capability. The most elegant design fails if managers cannot hold a straight conversation about a missed commitment. Ask honestly what proportion of your managers can, then design for that number rather than the one you wish you had. A simpler system run well beats a sophisticated one run badly, every time.
Sector and regulation. In safety-critical or heavily regulated work, documentation and consistency are not bureaucracy, they are the job. In a fast-moving commercial business, the same weight of process buys nothing.
How interdependent the work is. Where results come from teams, individual ratings will misattribute both credit and blame, and the more interdependent the work, the worse it gets.
Growth stage. A company doubling in size needs standards that transfer to managers who have never met each other. A stable one can rely on judgment built over years.
The labor market you hire in. If your people can leave on Friday and start elsewhere on Monday, a system that feels arbitrary is a retention problem, not just a fairness one.
Write down where you sit on all five before anyone proposes a framework. Half the disagreements in a redesign meeting are people arguing from different answers to these questions without saying so.
Culture decides what survives contact
A system only works if the managers in your organization will actually run it. That is not a soft consideration, it is the main one.
Research on why performance management stays broken makes exactly this point 3: the formal system gets reduced to prescribed administrative steps that are disconnected from the day-to-day activities where performance is actually managed 2. A beautifully designed quarterly process, run by managers who avoid difficult conversations, produces documentation and nothing else.
So before choosing, ask what your managers can carry. If they cannot have a straight conversation about a missed commitment today, no framework will supply that. Fix the conversation, or design a system that needs less of it and says so honestly.
Give the current practice a fair hearing
Before removing something, name the useful work it does. A recurring check-in may be the only reliable time an employee has with their manager. A short annual record may be the only context that survives a manager moving on. A calibration meeting may be the only place cross-team work is seen.
Retain when the purpose is clear and the cost is acceptable. Redesign when the purpose matters and the method keeps failing. Remove when it duplicates something better or creates problems a simpler approach avoids. The same organization can reasonably do all three in different places, and that is a more honest answer than a single new model.
Decide what you want to be true in two years
The most useful question in the room is not about the process at all. In two or three years, what should be different that anyone outside HR would notice?
More people attempting harder work. Fewer good people leaving for reasons you could have fixed. Pay decisions that managers can explain without flinching. Problems reaching the people who can solve them inside a month.
Pick one. Then choose the smallest change that could move it, name who owns it, set the date you will look again, and write down in advance what would tell you it is not working. If nothing would make you reverse it, you are not running a test. You are announcing a policy.
What the evidence supports, and how far
Effectiveness has several separate criteria and has to be judged against the decision being made 1. Everyday practice matters more than the design of the form, and formal systems drift into prescribed steps disconnected from the work 2, 3. Whether a decision can be explained to the person it affects has a large literature of its own 4.
The outcome categories, the gap table and the two-year question are ours, built from the preceding articles. They are a way to structure the decision, not a validated diagnostic.
Evidence: contested for the practices, proposition for this guide.
Our position
Pick one decision that has to get better, change the smallest thing that could move it, and write down in advance what would make you reverse it. Redesigns that begin with the form end with a new form.
Take it into the room: the Retain, redesign, remove worksheet turns this article into one printed page of questions, a framework and a table to fill in.
What would change this conclusion
Results from organizations that ran a small test and recorded what they decided afterwards. If teams working through this sequence reach clearer decisions than teams that redesign the form first, the guide earns its place. If the same disagreements survive the exercise, the constraint sits earlier in the chain, usually in who reconciles commitments that conflict.
Read next: Do individual goals add up to organizational performance?
Evidence map
- Schleicher and colleagues (2019), Evaluating the Effectiveness of Performance Management. Integrative conceptual review; indexed abstract accessed.
- Pulakos, Mueller-Hanson and Arad (2019), The Evolution of Performance Management. Research review; relevant full text accessed.
- Pulakos and O'Leary (2011), Why Is Performance Management Broken? Industrial and Organizational Psychology. Published abstract accessed in full: formal systems become prescribed steps disconnected from the day-to-day activities that determine performance.
- Colquitt, Conlon, Wesson and colleagues (2001), Justice at the millennium. Journal of Applied Psychology. Record and title verified; abstract not retrievable through the sources checked. Cited for the existence and scale of the fair-process literature.
Sources checked September 22, 2026. Prepared with AI assistance.