OKRs without the cult

OKRs work when you use them to align a small number of things that matter. They fail when you use them as a reporting framework, a performance review tool, or a theater of strategy. Most OKR implementations I've seen are the latter, and the teams running them would be measurably better off without them.

What OKRs actually are

• Objective. A qualitative, directional statement. What you want to be true by the end of the period. One sentence.
• Key Results. 2, 4 measurable outcomes that tell you whether the objective was achieved. Numbers with deadlines.

Objective = where we're going. Key results = how we'll know we got there.

The rules that keep OKRs useful

1. Max 3 objectives per team, per quarter. Teams with 7 "priorities" have 0 priorities.
2. Key results are outcomes, not outputs. "Launch feature X" is not a KR, it's a task. "Increase activation rate from 40% to 55%" is a KR.
3. Cascade by negotiation, not decree. The CEO's objectives don't automatically become the VP's objectives. Each level chooses what they'll contribute.
4. Not tied to comp. The moment bonuses depend on hitting KRs, people sandbag targets. OKRs become commitments instead of stretches.
5. Review weekly, grade at end. Weekly: "where are we, what's off-track, what do we need?" End-of-quarter: "did we hit it? If not, why?"

How to write a good KR

A good KR passes three tests:

• Measurable. A number with a source of truth. "Improve customer satisfaction" fails. "NPS from 32 to 45, measured by quarterly survey" passes.
• Time-bound. By when? Usually end of quarter.
• Causally connected. Achieving this KR would actually move the objective. (If you hit the KR and the objective still isn't achieved, you wrote the wrong KR.)

Good OKR example.
Objective: Make the onboarding experience so good that new customers get to value in the first session.
KR1: Time-to-first-key-action: 18 min → 7 min
KR2: % of new signups who complete onboarding: 52% → 72%
KR3: Week-4 retention: 31% → 42%

Ambitious vs committed

Google popularized "moonshot" OKRs where grading 0.7/1.0 is a win. In practice, in real businesses:

• Committed OKRs, must hit. Usually revenue, customer-facing commitments, compliance. Graded binary: did / didn't.
• Aspirational OKRs, try hard; 0.7/1.0 is good. Usually new initiatives, exploratory work, stretch goals.

Mark which is which when you set them. The team needs to know which numbers they must hit and which ones are stretch.

The most common failure modes

OKR theater

Teams write OKRs, post them, never reference them again. The quarterly OKR review feels like a chore because nothing actually depends on it. Fix: link the weekly business review to OKR progress. Make it the first agenda item.

Top-down cascade

Executives write OKRs, then assign them down. Teams treat them like tasks, not goals. No ownership. Fix: executives set company-level objectives; teams choose their own contributing OKRs.

Too many OKRs

10 objectives per team, 5 KRs per objective = 50 tracked numbers. Nobody can hold that in their head. Fix: 3 objectives max, 2, 4 KRs each.

OKRs become a task list

KRs read like: "Ship feature X. Run campaign Y. Hire 3 people." These are projects, not outcomes. Fix: for every KR that sounds like a task, ask "which metric would improve if we did this?" That's the real KR.

What good looks like

• Every person on the team can recite the current quarter's company OKRs
• Weekly reviews reference progress on KRs, not just activity
• When new work arrives, it's evaluated against OKRs ("does this advance an objective, or are we just doing it?")
• Quarterly grading actually happens, and misses become inputs to the next quarter's planning

Related: Annual + quarterly planning · What to kill · Weekly business review

What to do with this

• Cap OKRs at 3-5 per team, more than that fragments focus and trains the team to ignore half
• Write KRs as measurable outcomes, activity-based KRs ("launch feature X") miss the point
• Aim for 70% achievement, hitting 100% means targets were sandbagged, missing 50% means disconnected from reality
• Review OKRs weekly in the business review, OKRs that aren't in weekly operating cadence become documents
• Kill OKRs if the whole system becomes theater, cargo-cult OKRs are worse than no system