Running OKRs
Setting OKRs is the easy part. Running them through a quarter is where most teams lose the thread. A predictable rhythm, a short weekly format, and an honest end-of-cycle grade are what keep OKRs from becoming planning theatre.
The cadence
OKRs are run, not set and forgotten. A predictable rhythm does the work: a light weekly pulse, a fuller monthly look, a quarterly reset. The rhythm matters more than the ceremony. Shorter loops learn faster. Skipping the cadence is the most common way OKRs quietly die.
Ask first: "Who is confirming the recurring calendar invites before we write the first Objective?"
- Get the cadence into calendars before the first Objective is written, not after. The order matters.
- Get the sponsor to commit to attending, not just approving. A cadence the senior person skips is a cadence the team will skip.
- If a standing slot feels impossible, agree on a lighter version: a shared async update with a monthly face-to-face. Something beats nothing.
Example: cadence that survives a busy month
"We will check in when we can" is the agreement. By week four, it has never happened. The OKRs sit in a shared doc nobody opens.
A standing 15-minute slot every Monday, a 45-minute monthly review, both in the calendar for the full quarter, both with the sponsor in the invite. When a busy week hits, the slot is already there.
Transparency only exists if the looking-at-it is a habit. An intention to check in is not a habit. A recurring invite with a senior owner is.
Transparency only exists if the looking-at-it is a habit, not an intention.
The weekly check-in
The weekly is a short pulse, not a status meeting. Three questions per Key Result: how confident are we will hit it (say 1 to 5), what moved it this week, and what one thing would move it most next week. Confidence is the signal: a stretch KR sitting at very high confidence in week one is probably a task. Everything at low confidence is a wish.
Ask: "What is your confidence on this, 1 to 5, and what is it based on?"
- In the early weeks, write up the answers yourself rather than waiting for a template someone fills in alone. It builds the habit faster.
- When confidence is high but the number is not moving, push on what the confidence is based on. Confidence without evidence is hope.
- Keep it to 15 minutes. If it runs longer, the format is wrong, not the time limit.
Example: a 15-minute weekly
A 60-minute status read-around. Each person reports what they did. Nothing is decided. Nobody raises a concern because there is no structured moment to do it.
Per Key Result: "confidence 3, support tickets dropped slightly this week, next week we ship the in-app hint that should close the gap." Done in 15 minutes. One KR flagged amber, action agreed before the meeting ends.
Transparency is cheap when it is small and frequent, expensive when it is rare and heavy. The 15-minute format forces the question that matters: what one thing would actually move this?
Transparency is cheap when it is small and frequent, expensive when it is rare and heavy.
Grading and learning
At the end of the cycle you grade, not to judge people but to learn. For a beginner team, keep it simple: red, amber, green plus a one-line "what we learned", and add a finer scale later. Normalise that a stretch goal landing at partial is a good result, not a failure. The grade is the start of the next conversation, not a verdict.
Ask mid-cycle: "If today were the last day of the quarter, how would you grade each KR, and what would you say you learned?"
- Rehearse grading at mid-cycle, as if today were the last day. The political discomfort that surfaces is the real content. Deal with it before the grade is real.
- Separate aspirational goals (partial is fine) from committed ones (must hit) before any number appears. If the team does not know which is which, that is the first problem to fix.
- When a KR lands amber, ask "what did we learn?" before asking "what went wrong?". The order changes the conversation.
Example: a "failed" KR that was a win
A KR grades amber. The team braces for blame. The review becomes a defence of decisions made three months ago. Nothing useful comes out of it.
Same amber grade, but the conversation starts with "what did we learn?" The team surfaces that the real blocker was pricing, not onboarding, which reshapes the next quarter's Objective before anyone has left the room.
Learning out loud is alignment for the next cycle. But only if the grade is safe to be honest about. The mid-cycle rehearsal is what makes the end-of-cycle honest.
Learning out loud is alignment for the next cycle, if the grade is safe to be honest about.
Paste an OKR into Diagnose and get a score in 60 seconds.