Teams switching to OKRs often talk as if KPIs are the old, boring way and OKRs the modern replacement. That framing is wrong, and it makes companies adopt a fashionable system without understanding what problem it solves. These are two different instruments. Let me lay out each, compare them honestly, and say where I land.
What a KPI does
A key performance indicator is a metric that tells you whether something you must keep running is running well. Call-centre first-call resolution, invoice processing time, monthly recurring revenue, bug escape rate, offer acceptance rate. It has a number, a target, a measuring period, and usually it stays on the dashboard year after year. KPIs describe the health of the ongoing business. If a KPI moves out of range, someone investigates.
The strength is stability. You can compare this quarter with last year. The weakness is that a KPI says nothing about what to change. A team can hit every KPI and still be heading nowhere new.
What an OKR does
Objectives and Key Results is a goal-setting method associated with Andy Grove at Intel and spread widely through John Doerr’s book Measure What Matters, with Google among its best-known users. An objective is a qualitative direction, something worth achieving this quarter. Key results are a handful of measurable outcomes that show whether you got there. For example, an objective of “make onboarding painless for new customers” might carry key results such as cutting median setup time from nine days to four, and raising week-four activation from 55 to 70 percent. Those figures are illustrative.
OKRs are deliberately about change. They are set for a short cycle, commonly a quarter, they are meant to be visible across teams, and many companies expect ambitious “stretch” targets where hitting around 70 percent counts as a good result. That last convention is exactly where trouble starts, as we shall see.
Side by side
| KPIs | OKRs | |
|---|---|---|
| Main question | Is the ongoing work healthy? | What should we change or achieve next? |
| Time horizon | Continuous, reviewed monthly or quarterly | Short cycle, usually a quarter |
| Targets | Realistic, expected to be met | Often ambitious, partial achievement accepted |
| Count | Many, depending on role | Few, often three to five objectives |
| Typical failure | Gaming, tunnel vision | Vague objectives, activity dressed up as results |
How each one goes wrong
KPI failure is mostly Goodhart’s problem: once a measure becomes a target, people optimise the measure. If a support team is judged on average handling time, agents rush customers off the call. If recruiters are judged on time to fill alone, they push weaker candidates through. The number looks great and the real outcome suffers. The usual remedy is to pair metrics that pull against each other, for example speed with quality.
OKR failure is subtler. Companies write key results that are really tasks: “launch three webinars”, “conduct 20 interviews”. Those are things done, not results achieved. Or they set an objective so broad that any activity could count. And many organisations copy the cascading approach rigidly, so every team’s OKR is a smaller copy of the one above, and nobody has room to think.
The question of pay
This is where I take a position, and it is an opinion. Do not link ambitious OKRs directly to bonus or appraisal ratings. If an employee knows that missing the target costs them money, they will propose safe, sandbagged targets, and the entire “aim high, accept 70 percent” logic collapses. People protect their income before they chase a moonshot. Meanwhile, KPIs for stable operational roles can reasonably feed into variable pay, because their targets are set to be achievable, as with a sales incentive plan or a monthly processing target.
So my arrangement would look like this. Use KPIs to run the business and, where appropriate, to determine variable pay. Use OKRs for a small number of priority changes, reviewed in conversations and not auto-scored into ratings. A manager can still consider OKR progress in an appraisal as one piece of evidence, with context, which is different from a formula.
Some companies disagree and tie OKR scores to pay; if you do that, expect to set targets more conservatively and to accept that the stretch idea is mostly gone. Be honest about which system you are running.
When to use which
A back-office payroll team with stable processes, well-understood volumes and clear service levels is better off with KPIs. Adding OKRs may produce theatre. A product team entering a new market, or an HR function rebuilding its hiring process, benefits from OKRs because the work is a shift, not a steady state. A company’s small size matters too. Under fifty people, with everyone in one room, a written list of three priorities may do the job without any framework.
Whichever you pick, check whether people can recite the goals without looking them up. If they cannot, the system has become paperwork. The simplest test of a goal framework is whether Tuesday’s decisions change because of it.
Figures and names in the examples are illustrative.
Leave a Reply