Somebody in your company has proposed that the team “move to OKRs.” Somebody else has pointed out that you already have KPIs. A third person has said that whatever you do, the goals should be SMART. Forty minutes later the meeting ends with an action item to “align on a framework,” which nobody does, and next quarter you set goals exactly the way you set them last quarter.
I have been in that meeting more times than I want to count, on both sides of it. It took me an embarrassingly long time to notice why it never resolves.
The question “should we use OKRs or KPIs?” is malformed. It sounds like a choice between two tools that do the same job, the way you would choose between Slack and Teams. They are not that. OKRs, KPIs and SMART goals answer three genuinely different questions, and a team that is working well usually has all three running at once, doing different work. Once you see which question each one answers, the argument dissolves, and the practical version of setting team goals as a new manager gets a lot simpler.
This is the comparison nobody gave me.
The three questions
Here is the whole article in one table. Everything below is detail.
| Answers the question | Time horizon | Changes? | |
|---|---|---|---|
| KPI | What should never get worse? | Ongoing, no end date | You want it stable or slowly improving |
| OKR | What are we trying to change this quarter? | One quarter, occasionally a year | Yes, that is the entire point |
| SMART | Is this goal written well enough to act on? | Not a horizon, a quality check | It is a standard, not a target |
The confusion comes from the fact that all three produce sentences with numbers in them. That surface similarity hides the difference in purpose, which is where every framework argument actually lives.
KPIs: the things you watch, not the things you chase
A key performance indicator is a number you monitor continuously because it tells you whether the machine is still running. Support response time. Monthly churn. Uptime. Gross margin. Error rate.
The defining property of a KPI is that it has no end date and often no target. It is a dial on the dashboard. You are not trying to finish it. You are trying to notice when it moves in a way that means something.
That property is exactly what makes KPIs valuable and exactly what makes them useless as goals. “Keep churn under 3%” is not something the team achieves in March and then celebrates. It is a condition you maintain forever, which means it generates no urgency, no sequencing, and no decisions about what to do first.
Where KPIs go wrong for small teams:
Measuring what is easy instead of what matters. Page views, emails sent, tickets closed. These are countable, which is not the same as informative. The test: if this number doubled tomorrow, would the business genuinely be better off? If you cannot answer confidently, it is a vanity metric wearing a KPI badge.
Nobody owns it. A KPI with no named owner gets reported and never acted on. Every dial needs one person whose job it is to say something when it moves.
Too many of them. A team of six with fourteen KPIs is not measuring performance, it is generating a report. Five or six is plenty at team level. Anything you would not act on within a week does not belong on the list.
Confusing the dial with the destination. This is the big one. When a KPI becomes the goal, people optimize the number rather than the thing it represents. Support times drop because tickets get closed prematurely. This failure is common enough to have a name, Goodhart’s law: a measure that becomes a target stops being a good measure.
OKRs: the change you are attempting
Objectives and Key Results, popularized by John Doerr in Measure What Matters after he brought the practice from Intel to Google, answer a different question. Not “what should stay healthy,” but “what are we deliberately trying to make different by the end of this quarter?”
The structure is two-part on purpose:
- The Objective is qualitative and directional. It should be memorable and slightly uncomfortable. “Make onboarding something customers finish without us.”
- The Key Results are the two to four measurable outcomes that would prove the objective happened. Not the tasks you will do. The results those tasks would produce.
That last distinction is where most teams fall over, so it is worth being blunt. “Rewrite the onboarding docs” is a task. “Cut the median time to first successful use from 9 days to 3” is a key result. If your key results are a to-do list with checkboxes, you have written a project plan and labelled it an OKR, and you will get the overhead of the framework with none of the benefit.
About the ambition convention. In OKR practice, particularly the version that came out of Google, teams are encouraged to set key results ambitiously enough that hitting roughly 60 to 70 percent counts as success. Two honest caveats. First, this is a practitioner convention, not a research finding, and you will see it quoted as though it were the latter. Second, it only works if it is genuinely safe to miss. In a company where 70 percent attainment shows up as underperformance in a review, everyone learns within one cycle to set key results they can definitely hit, and you have rebuilt ordinary targets with extra ceremony. If you cannot promise that safety, do not import the ambition norm. Set OKRs you intend to hit and be honest about it.
Where OKRs go wrong for small teams:
Too many. Three objectives is a lot for a small team. Five is a wish list. The whole value of the format is forcing a choice about what matters most this quarter, and that value is destroyed the moment you avoid the choice by listing everything.
Cascading them too literally. Cascading is supposed to mean each level can see how their work connects upward. It is not supposed to mean every person gets a personal miniature of the company OKR. When a five person team has four levels of cascade, the paperwork exceeds the work.
Setting them and not looking at them. By far the most common failure, and not really an OKR problem at all. It is a rhythm problem, and it kills SMART goals just as efficiently.
SMART: a writing standard, not a framework
This is the reclassification that makes the whole comparison make sense.
SMART (Specific, Measurable, Achievable, Relevant, Time-bound) does not tell you what to pursue. It tells you whether a goal you have already chosen is written clearly enough to be acted on and assessed. It is a quality check, and you can run it on an OKR key result, on a personal development goal, or on a commitment you made to your boss.
Which means “should we use SMART goals or OKRs?” is roughly like asking whether you should use grammar or paragraphs.
What makes SMART worth keeping is that two of its five letters rest on unusually solid ground. Edwin Locke and Gary Latham spent decades testing what actually makes goals work, summarized in their 35-year review in American Psychologist, and the central finding is consistent and heavily replicated: specific and difficult goals produce higher performance than vague goals or “do your best.” By 1990 the effect had been found across 88 different tasks and more than 40,000 participants, in laboratory and field settings, across countries, with time spans from one minute to 25 years. It is one of the most validated findings in organizational psychology.
Note the word difficult. Goal-setting theory does not say achievable goals work best. It says specific and hard ones do, with two conditions attached: the person has to be committed to the goal, and they have to get feedback on progress. That is an uncomfortable fit with the A in SMART, which most people read as “safe.” A better reading of A is “possible with real effort,” not “comfortable.”
The two conditions matter more than the acronym. Commitment usually comes from having been involved in setting the goal rather than receiving it. Feedback comes from the review rhythm, which is the subject of the next section, and which is the actual reason most goal systems fail regardless of which letters you use.
How the three stack
In a team that is working, the layers look like this:
Layer 1, KPIs. Five or six numbers that describe the health of what you run. Reviewed monthly, or weekly if they move fast. Mostly boring, and boring is the point.
Layer 2, OKRs. One to three things you are trying to change this quarter, each with two to four key results. Reviewed weekly, in fifteen minutes.
Layer 3, SMART. Applied to the wording of every key result and every individual goal, so that nobody has to guess what “improve onboarding” means.
The relationship between layers one and two is the useful part. Your OKRs usually come from a KPI that is not where you want it, or from something you want to build that no current KPI covers. Churn drifting up is a dial. Deciding that this quarter you will fix the two biggest drivers of it is an objective. And when the OKR succeeds, the improvement it produced becomes the new normal that the KPI then protects.
That loop is the whole system. Dials tell you where to look. OKRs are how you change something you did not like. SMART is how you write both so people can act on them.
Which one should you actually start with?
For a first-time manager, in order:
| Your situation | Start with | Why |
|---|---|---|
| No goals at all, team of any size | Three SMART team goals | You need the habit before the framework. Frameworks fail on teams with no goal rhythm |
| You have a working goal habit, want more ambition | OKRs, one objective | Add ambition to a rhythm that already exists, one objective before three |
| You inherited a dashboard nobody acts on | Prune the KPIs first | Cut to five you would act on. Then choose one to attack with an OKR |
| Company mandates OKRs, you are not ready | Write OKRs, run SMART underneath | Satisfy the format upward, keep the clarity that actually helps your team |
The general advice from the team goals guide holds here: do not start with OKRs because a famous company uses them. Start with three clear goals and a weekly fifteen minute review. If you can hold that for a full quarter, adding OKRs is easy. If you cannot, OKRs will not save you, they will just give you a more sophisticated way to fail.
The mistake that kills all three
Every framework on this page dies the same death, and it has nothing to do with the framework.
The planning gets four to eight hours. The maintenance gets zero. A goal reviewed once at the end of the quarter is not a goal, it is a prediction you get graded on, and in the eleven weeks between, it competes with the actual week and loses every time.
Locke and Latham’s own conditions say it plainly: goals need commitment and feedback. A quarterly review provides feedback once, far too late to change anything. Work reported by MIT Sloan Management Review points in the same direction, finding that goal systems built around frequent, transparent updates outperform those built around getting the specification right up front. The fifteen minute weekly check is not administrative overhead on top of the goal system. It is the goal system. Everything else is stationery.
Three questions, once a week, same slot: what did we say we would move, what actually moved, what is in the way.
Getting the mechanics right
Two practical resources, because at some point you have to stop comparing frameworks and write the thing.
For the mechanics of writing OKRs and SMART goals well, including worked examples by function and copy-paste templates, Celorly’s knowledge hub covers the step by step versions: how to write key results that are outcomes rather than tasks, what good goals look like for marketing, sales, product or HR teams, and how to run the check-in without turning it into a status meeting. It is aimed at small teams, which is the context most goal-setting advice ignores in favour of enterprise rollouts.
For the thinking behind the frameworks, Doerr’s Measure What Matters is the canonical OKR source and Locke and Latham are the research foundation underneath all of it. Both are covered in our five books on goal setting and getting results, along with what each one is actually good for.
What to do this week
- Write down the five or six numbers that tell you whether your team is healthy. That is your KPI list. Delete anything you would not act on.
- Pick the one you least like the look of, or the one thing you want to build that no number currently covers. That is your objective for the quarter.
- Write two to four key results for it. Check each one: is this an outcome, or is it a task with a date? Rewrite the tasks.
- Run each through SMART as a writing check, and be honest about the A. Possible with real effort, not comfortable.
- Put fifteen minutes in the calendar, weekly, same slot, starting this week. This step is the one that decides whether any of the above survives past week three.
If you want a read on whether your current goals are set up to survive the quarter before you rebuild anything, the Are Your Team Goals Setting You Up to Fail? assessment walks through the specific failure patterns in about three minutes. Most of them are rhythm problems wearing a framework costume, which is the theme of this entire article.
The framework argument is not worth forty minutes. The weekly fifteen is worth all of them.