Less than the rollout promises. What changes performance sits underneath every label, and which format helps you depends on your context, not on which one is fashionable.
For the CHRO, HR leader or business leader who owns the performance system. Article 2 of seven in the performance management investigation.
You are asked to rewrite your goals because the company is moving to OKRs. Last quarter, reducing repeat complaints was a KPI. This quarter it sits under an objective about customer trust. Your team is still waiting for operations to fix the delivery problem.
Something did change. The paperwork. Whether anything else changed depends on questions the rollout rarely asks.
Four things people confuse
Most arguments about frameworks dissolve once these are separated.
| Word | What it actually is | Example |
|---|---|---|
| Responsibility | Work the role owns all year | Handle customer complaints well |
| Measure | What you watch to tell how it is going | Repeat complaints per hundred orders |
| Target | The number you are committing to | Down from twelve to six by March |
| Expectation | What good work looks like here | Customers should not have to ask twice |
For this article, we use a working distinction: a KRA describes a responsibility, a KPI a measure, a target a commitment, and an OKR an objective with key results. Organizations use these labels differently. None of them is a substitute for the expectation, which is the part employees actually ask about and the part most systems leave unwritten.
What each side needs
Before choosing a format, be clear about whose problem you are solving.
The employee needs clarity: what to focus on now, over what time frame, and what counts as done well. Most frustration attributed to "the OKR process" is really a person carrying eleven priorities and no ranking.
The manager needs a simpler conversation: something that makes it easy to agree priorities, to change them when the business changes, and to hold someone to account without the exchange turning personal. A format that takes three hours a quarter to administer fails this test regardless of its logic.
The organization needs ambition where ambition pays: goals that stretch a team into work it would not otherwise attempt, in the places where attempting and missing is acceptable. In other places, reliability matters more than stretch, and a stretch goal is an invitation to break something.
Those three needs are compatible. They are just rarely written down before a framework is picked.
What actually changes performance
Four things, and they sit under every label: the goal is specific, it is hard enough to require real effort, the person is committed to it, and feedback arrives while the work can still change 1.
The evidence for those four is strong. The evidence that one framework beats another is not. The most useful study available on objectives and key results is interview and survey work inside a single company about how teams adopt them 2. It describes implementation difficulty. It does not show superiority over anything.
That is worth saying out loud during a rollout, because the claim being made from the stage is exactly the claim nobody has tested.
Context decides
If the ingredients are constant, what should drive the choice? Your situation. What follows is our judgment as practitioners, not a tested rule.
Plain expectations suit small teams and stable service work, where what good looks like is visible and the manager is close to the work. Adding a framework adds admin and subtracts nothing.
KPIs and KRAs suit operational work with clear cause and effect, where the measure genuinely reflects the job and the target can be set honestly. Their failure mode is the measure drifting away from the work it was chosen to represent.
OKRs suit ambiguity: cross-team ambition, a quarterly rhythm, work where the right path is not known at the start. Their failure mode is ceremony, and the tell is a company where every objective is graded and nothing is abandoned.
So the honest advice for someone considering a switch is not "yes" or "no". It is: describe the difficulty you are having in one sentence, and check whether a format is even capable of fixing it. A delivery dependency nobody owns is not a goal format problem, and moving it under a shiny objective will not move it an inch.
Before you switch
Three questions, in this order.
What outcome do we actually want? More ambition, clearer focus, better conversations, faster reprioritization. Each points at a different change, and only one of them is about the label.
Could we get that by writing the existing goals better? Usually, yes. Most of the value people attribute to a new framework comes from finally naming a single priority, agreeing what good looks like, and setting a date to revisit it. You can do that on the current form this week.
How could someone hit this number without doing the work? When a measure carries consequences, people tend to work on the measure rather than the thing it stands for. Accounting research calls it surrogation 3, and every label inherits it. Repeat complaints can fall because the problem was fixed, or because tickets are being closed and reopened under new numbers.
What the evidence supports, and how far
The ingredients that change performance are well supported 1. Framework comparison is not: the available research on objectives and key results covers adoption inside one company 2, and the widely circulated playbook describes a practice rather than testing it 4. Surrogation is an experimental finding about measures that carry consequences 3.
The fit-to-context guidance is ours, drawn from practice rather than demonstrated by a study.
Evidence: established for the four ingredients, not yet established for any claim that one framework beats another.
Our position
Do not change the label. Write the existing goals better: one priority, what good looks like in a sentence, and a date to revisit. If you still want a new framework after that, you will at least know what you are buying.
Take it into the room: the Say what you actually mean worksheet turns this article into one printed page of questions, a framework and a table to fill in.
What would change this conclusion
Comparable teams running different frameworks with the quality of the underlying conversation held constant. Or evidence that a particular format reliably produces more specific goals, firmer commitment or faster reprioritization, which would be the label working through the ingredients rather than beside them.
Read next: Does more frequent feedback produce better work?
Evidence map
- Locke and Latham (2002), Building a Practically Useful Theory of Goal Setting and Task Motivation. Research synthesis; relevant full text accessed.
- Butler, Zimmermann and Bird (2023), Objectives and Key Results in Software Teams. Authors' paper; abstract accessed. Interviews and survey responses inside one company. Illuminates implementation difficulty, establishes no comparison between frameworks.
- Choi, Hecht and Tayler (2009), Lost in Translation: The Effects of Incentive Compensation on Strategy Surrogation. Experimental accounting research. Bibliographic record verified; working paper version, full text not accessed.
- Google's OKR Playbook, reproduced with permission by What Matters. Practitioner guidance; relevant sections accessed. Describes one company's practice.
Sources checked September 22, 2026. Prepared with AI assistance.