Seven steps to run OKRs with a team

Identify the real sponsor: the person who controls the team's calendar, not the one who approved the idea. Gather whatever the team currently calls its goals and note any existing metrics. Look up the team's last quarter to spot recurring failure modes.

• Find who funds the team's headcount. That person knows who owns the outcome.
• Ask directly: "If this engagement fails in three months, who is on the hook?" Their answer is the sponsor.
• Do not start the readiness conversation until you have a named individual, not a committee.
• Prepare a one-line framing of the cost of no sponsor before the call.

• Map who holds which type of authority: budget, calendar access, public accountability.
• Write a one-page note that names the risk: when two people share sponsorship, neither feels the cadence is theirs to protect.
• Propose a single "ritual owner" who carries the cadence even if strategic calls stay shared.
• Bring the note to a pre-meeting with the most senior of the leaders before the full conversation.

• Open with the why: what focus and transparency would change for this team this quarter.
• Name the time cost plainly: one 15-minute weekly, one 45-minute monthly, one 90-minute quarterly review.
• Ask the sponsor to model attendance, not just approve it.
• If they will not commit to sending the calendar invites before Step 1, surface it now, not later.

• Validate the scepticism: name a past goal-setting attempt that quietly faded and ask what made it fade.
• Offer a smaller commitment: agree the cadence for one cycle only, with an explicit kill criterion.
• Ask for two attendance commitments, not all four: weekly and quarterly. Skip the monthly until trust builds.
• Walk away if the sponsor will not commit to either. A sceptic who attends is fine; a sceptic who skips kills the rhythm.

• Pre-read names the time cost, the sponsor's role, and the one decision you need: cadence sign-off.
• Use the call only for the commitment. Do not re-explain OKRs.
• Get the calendar invites sent from the sponsor's account during the call, not after.
• If you cannot get the call, escalate to written commitment with two reply-by dates.

Send the calendar series yourself if the sponsor drags their feet, with a note asking them to forward it to the team. Tell the team that coaching starts and ask them to bring their current goals to Step 1.

Homework for the team

Sponsor sends the recurring invites. Team members note down their current goals to bring to Step 1.

• Make explicit: the delegate coordinates logistics; the sponsor still models attendance.
• The delegate substituting for the sponsor in weekly check-ins is not acceptable after the first two.
• Set a named check-in date with the sponsor directly at week 3.
• Confirm invites are sent from the sponsor's calendar, not the delegate's.

• Frame it as a rhythm check, not a performance review: "Is this working for you so far?"
• Use it to test whether the cadence holds or whether to renegotiate scope.
• Prepare a one-page "what we agreed" note to anchor the conversation.
• If the rhythm has already slipped, offer to reduce scope for the first cycle rather than accelerate.

Write each of the team's current goals onto cards before the session, one goal per card. Set up three columns labelled Output, Outcome, and Impact. Have the cards and columns ready before anyone arrives.

• Ask the team what they remember about the last goal-setting exercise and why it stopped mattering.
• Write the reasons on a visible surface. Do not defend the past approach.
• Name the one or two things this approach does differently. Keep it brief.
• Then move directly into the sort. The retro is not the session; it is the warm-up.

• Ask each person to write: what do I own, what do I depend on, and what do others depend on me for?
• Share in a brief round. Two minutes per person maximum.
• Identify any goal cards that belong to a single person versus the team as a whole before sorting starts.
• Then proceed with the standard sort.

• Write each current goal on its own card, one per item.
• Everyone sorts silently. Disagreement during sorting muddies the result.
• Reveal the piles. Discuss the disagreements: disagreement is the learning, not a problem to resolve.
• Put three questions on the wall: are we focused, are we transparent, are we aligned?

• Silent sort first: everyone places their cards independently with no commentary.
• Dot-vote on contested placements before any plenary discussion.
• In the plenary, invite junior voices first. Ask the lead to wait until at least two others have spoken.
• If the lead's view diverges sharply from the team's, name it as a useful tension rather than resolving it immediately.

• Set up three columns on the shared board before the session and share the link in the invite.
• First 5 minutes: everyone drags their cards independently with cameras off.
• Next 10 minutes: random pairs share their reasoning in breakout rooms or direct messages.
• Plenary: focus only on cards where pairs disagreed. Skip consensus cards.

Leave the three-question wall visible in the team's workspace. Photograph the sorted cards and share the photo so it is findable during the quarter.

Homework for the team

Each person rewrites one of their goals as an outcome before Step 2. Bring it to the next session.

• Identify which disagreement carried the most energy and name it explicitly as a thread to follow up.
• Offer a 20-minute 1:1 to the person whose card was most contested.
• Do not summarise the disagreement as "healthy debate" in public. Acknowledge it is unresolved.
• If the conflict is between two specific people, meet them separately before Step 2, not together.

• Identify the one goal everyone placed in the same pile immediately.
• Ask the team to write one sentence about why they all agreed, without consulting each other.
• Share the sentences at the start of Step 2. Divergence there surfaces latent disagreement safely.
• Do not assume quick consensus means shared understanding. It often means shared avoidance.

Review the team's strategy and scope so you can spot an Objective that falls outside what the team can actually influence. Bring the homework from Step 1 (each person's rewritten outcome goal) as raw material for the obstacle write.

• Find the highest-level goal document the team is supposed to be contributing to.
• Read it together with the lead and note where the team's work is mentioned and where it is absent.
• Draft one sentence that says what the team's contribution is in plain language.
• Use that sentence as the anchor in the team session. It does not need to be perfect; it needs to exist.

• Write down the two competing strategic directions in plain sentences, not jargon.
• Send them to the sponsor with a specific question: which one does this team prioritise this quarter?
• Do not run the Objective session until you have a written answer, even a brief one.
• If the sponsor cannot decide, that is the risk to name in the session, not to resolve through facilitation.

• Ask everyone to write silently: the single biggest obstacle I would remove this quarter if I could.
• Cluster the cards. Find the theme that carries the most energy in the room.
• Pick the biggest obstacle together. Resist the urge to merge two into one sentence.
• Flip it: what is the positive state you want to be in when this obstacle is gone? Write it as one sentence, no numbers.
• If several Objective candidates remain, give the team three votes total, not per person. Force a choice.

• Individual silent write: two minutes. No exceptions including the lead.
• Pairs share with someone they would not normally choose.
• Groups of four consolidate to one or two strongest obstacles.
• Plenary: invite each group to share. Ask the lead to respond to what they heard, not to present their own view first.

• List all Objective candidates visibly. Give each person exactly three votes, not three per candidate.
• No trading, no campaigning. Thirty seconds of silent voting.
• The top candidate is the Objective. If there is a tie, the lead breaks it, but only after the team has voted.
• Park the runners-up explicitly: write them down, acknowledge them as future candidates, and move on.

Post the Objective where the team works every day. Sanity-check it with the sponsor in a separate brief conversation: does it land within scope?

Homework for the team

Live with the Objective for a few days. If it stops feeling true, flag it before Step 3 rather than working from a shaky foundation.

• Ask each person to write one sentence: does this Objective still feel like the right bet? Why or why not?
• Share async or in a quick stand-up. Look for drift in language, not just thumbs up or down.
• If more than one person has reframed the Objective in their head, return to the clustering step rather than proceeding.
• One revision is allowed before Step 3. After that, proceed with what you have.

• Acknowledge the second candidate is real and worth pursuing. Write it down visibly.
• Ask: "If we could only achieve one of these by end of quarter, which one matters more?" Let the sponsor answer.
• Name the cost of two: split attention, diluted Key Results, muddier grading at the end.
• Park the second with a named date to revisit: the end-of-cycle review.

Pull whatever baseline data is reachable before the session: current numbers from dashboards, support logs, or previous quarters. Have the result-or-activity sort cards ready. Gaps identified now become the first discovery Key Results.

• List every metric the team expects to use as a Key Result target.
• For each one, ask: does anyone know the current number? If not, write a discovery Key Result with a two-week deadline.
• Name one person per gap as the data owner. Not "the team" as an owner.
• Run the full session with placeholders for the unknowns. Do not defer the session until the data arrives.

• List each metric the team plans to use. Ask: where does it come from, who can pull it, and who is accountable for its accuracy?
• Write an owner next to each metric. If no one can own it, it is not a usable Key Result yet.
• Identify which metrics need a new reporting line or dashboard to track consistently.
• Bring the lineage map to the team session as a visible reference rather than rediscovering the gaps there.

• Draft Key Result candidates together: aim for four to six, then cut.
• Sort each on a card: result or activity? An activity can be completed just by doing the work.
• For each activity card, rewrite it as a result together. Do not throw it away; use it as the starting point.
• Check each Key Result is within the team's influence.
• Add baselines. If unknown, create a baseline-discovery Key Result.
• Set confidence at 50 to 70 percent for stretch Key Results.

• Do not let the second pass start until the team agrees the first-pass card is an activity, not a result.
• Ask the "so what" question for every card: "If you do this, what changes for someone who is not in this room?"
• When the team rewrites an activity as a result, celebrate the rewrite visibly. It reinforces the pattern.
• If the team insists a card is a result and you disagree, ask them to evidence it: "How would we measure whether this result happened?"

• For each lagging Key Result, ask: "What would we see move three weeks before the final number shifts?"
• Write that early signal as a companion Key Result or a sub-metric on the confidence board.
• Aim for at least one leading indicator in the set. If none exists, the team will only learn at the end.
• If a leading indicator cannot be named, the team does not yet understand the mechanism behind the lagging metric.

Assign a baseline-discovery owner for every data gap. Schedule the short follow-up session once those numbers are in, no later than one week out.

Homework for the team

Each Key Result owner finds the baseline and names the data source before the follow-up session. No baseline, no Key Result.

• Write the discovery Key Result with a specific date and a named owner, not "the team".
• Set the follow-up date now. Do not leave it floating.
• At the follow-up, if the baseline is still unknown, reduce the Key Result to the discovery task only and defer the stretch target.
• Do not set targets against unknown baselines. A guess is not a stretch goal.

• Ask the team: "If you are 95% confident, what are you actually committing to rather than stretching toward?"
• Separate aspirational Key Results (aim to hit, expect to learn) from committed ones (must hit, non-negotiable).
• For any Key Result above 80% confidence, either raise the target or mark it as committed, not aspirational.
• A set with no Key Results below 70% confidence is a set the team expects to complete, not a set they are reaching for.

Set up the confidence board before the first weekly: a simple shared document or board where each Key Result gets its own row with confidence score, movement note, and next bet. Confirm the recurring weekly is in every team member's calendar.

• Identify who removed the slot from the calendar and whether it was a decision or a scheduling accident.
• Go to the sponsor with a specific ask: one 15-minute protected slot, same time each week, for the rest of the quarter.
• If the sponsor will not protect it, propose async-only: each owner posts their three-question update by a named time each week.
• At the two-week mark, check whether async is producing real updates or polite non-updates. Use the answer to revisit the live slot.

• Map the team's time zones and find two or three slot options that give each person a reasonable hour at least once per month.
• Publish the rotation at the start of the quarter so no one is surprised.
• For weeks when the slot is unreachable for some members, record a five-minute summary and post it in the team channel.
• Mark which updates came from async versus live. If async dominates, the cadence has already become optional.

• Confidence we hit this Key Result by end of quarter: 1 to 5. Write the number, do not talk around it.
• What moved it this week? If nothing moved it, say so. Silence is information.
• The one thing that would move it most next week. One thing, not a list.
• The coach writes up the summary in the early weeks. This models the format and lowers the admin barrier.
• Blockers that cross Key Results get noted, not solved in the room.

• After the first absence, check whether it was a one-off or a pattern. Do nothing publicly yet.
• After the second absence, contact the sponsor: "The lead has missed the last two weeklies. This risks the cadence."
• Do not soften the message. A lead who skips signals to the team that the check-in is optional.
• If the sponsor is the lead, go to the lead directly and name the attendance pattern without hedging.

• A confidence score that does not move is not being updated honestly or the team cannot see what would move it.
• Drop the 1-5 scale for two weeks. Replace it with a concrete question: "What single thing, if it happened this week, would make you more confident?"
• Write the answers down. After two weeks, check whether the named things happened.
• If they happened and confidence still did not move, the Key Result target may need revising.

The coach writes up the confidence board in the first few weeks. Nudge the one next bet per owner by direct message so it does not disappear into the week.

Homework for the team

Each owner does their one highest-leverage thing before next week. One thing only.

• At the end of the weekly, read out each Key Result and ask: "Who is the single person accountable for movement on this?"
• If the answer is "the team" or "everyone", that is the problem to name.
• Write the name next to the Key Result in the confidence board. Not a list of contributors; one name.
• At the next weekly, start with the named person, not a round-table.

• Identify the specific Key Result where numbers are moving but the team does not feel closer to the Objective.
• Hypothesise: is the metric wrong, is the target wrong, or is the Objective itself off?
• Use the session to decide one of: keep and trust the lag, revise the metric, or revise the Objective.
• Record the decision in the confidence board as a note, not just a number update.

Pull the Key Result movement data ahead of both sessions. Pre-mark which OKRs are aspirational versus committed so the team has a shared reference before any grade lands.

• Collect whatever movement data exists, even if it is incomplete or approximate.
• For each Key Result with no data, write "data gap" next to it rather than leaving it blank.
• Frame the gaps in the session as a process finding: "We set a Key Result we could not measure. That is something to fix next quarter."
• Do not skip the grading session because the data is incomplete. An approximate grade is more useful than no grade.

• Meet the lead one day before the grading session. Walk through each Key Result and agree on a preliminary grade.
• Where you expect a contested grade, agree on how to frame it: as a learning, not a verdict.
• Ask the lead to avoid using performance language ("we underperformed", "we failed") in the grading session.
• If the lead is likely to defend a green grade the data does not support, name that risk in the private session rather than in the room.

• Mid-cycle: ask each person to grade every Key Result red / amber / green as if the quarter ended right now.
• Normalise the result: a stretch goal sitting at amber with six weeks to go is a good result, not a failure.
• Use the mid-cycle grade to reset second-half bets, not to reassure.
• End of cycle: grade red / amber / green with a single "what we learned" sentence per Key Result.
• For each Key Result: decide recommit, adjust, or replace. Do not carry stale Key Results into the next quarter.

• Name the purpose explicitly: "This is not a performance review. It is the learning step that makes the next quarter better."
• Ask for 15 minutes only. If the sponsor agrees to 15, start the clock and keep it.
• Focus the 15 minutes on the one-line learnings only, not the full grading protocol.
• If the sponsor still resists, ask what they are worried about. The answer is usually about optics, and that is a coaching conversation to have, not to avoid.

• For each green grade, ask: "What is the data point that supports this?"
• If the team cannot produce a number or a concrete observation, record the grade as amber.
• Do not make this punitive. Frame it as: "We need to be able to share this grade with leadership. What would you show them?"
• After two or three evidence requests, the team usually starts self-policing. Let that happen.

Feed the learnings directly into the next cycle's Objective conversation. Share the grades openly with leadership so the learning is not kept private.

Homework for the team

Each owner writes their one-line "what we learned" before the end-of-cycle session. This is the most valuable output of the whole quarter.

• After the grading session, read each one-line learning aloud and ask: "Does this point toward an Objective for next quarter?"
• Write any that do as candidate Objectives on a visible surface.
• Email the list to the team and the sponsor within 24 hours of the session, not at the start of next quarter.
• Name who is responsible for bringing the candidates to Step 2 of the next cycle.

• Before the grading session ends, open the calendar and find a date within seven days for the next Step 1.
• Send the invite before anyone leaves the room or the call.
• Share the graded Key Results and learnings as pre-read for the next Step 1, not as a separate document.
• If the team needs a rest week, schedule it explicitly. An unscheduled gap becomes a permanent one.

Know the 12 anti-patterns well enough to ask the right question live, without consulting the reference. You are not naming a label; you are noticing a shape and reflecting it back. Read the full catalogue before Step 2.

• Review the full catalogue and identify the three patterns that appeared most in the team's Step 1 goals.
• For each of those three, memorise one question you would ask if you spotted it live.
• Keep a physical or digital one-card reference for the first four weeks. Do not apologise for using it.
• By week five, retire the cheat sheet. You should recognise the patterns from the team's own language by then.

• If the team says "that is a vanity metric" before you do, ask: "What question would you ask to test that?" Push the thinking further rather than agreeing.
• When a team member names a pattern correctly, acknowledge it briefly and move to the question: "Yes. So what would you rewrite it as?"
• Avoid pattern-labelling becoming a way for experienced members to shut down ideas from newer ones.
• If the catalogue vocabulary is being used to criticise rather than improve, name that as its own pattern.

• For an output dressed as a Key Result: "If you ship that, what changes for a user?"
• For a vanity metric: "If that number doubles, who is better off and how would we know?"
• For a planning-theatre Key Result: "Is there any scenario where this could fail to move the needle even if you do everything right?"
• Hand the rewrite back to the team. Do not rewrite it for them.
• One pattern at a time. Naming three in a single session overwhelms.

• Ask the question as if it applies to any Key Result of that type, not specifically to the lead's.
• "For any Key Result where we track page views, what would have to change for that number to matter?" applies to everyone in the room.
• If the lead defends it, hold the question open: "Fair point. What evidence would convince you it is measuring the right thing?"
• Do not drop the question to avoid discomfort. A lead whose metric escapes scrutiny teaches the team that scrutiny is optional.

• Book a 15-minute conversation with the senior person before the next team session.
• Name the pattern directly in private: "I noticed the Key Result looks more like an activity. Can we talk about that before the team session?"
• Agree on the rewrite together. The senior person then introduces it in the team session as a revision they made.
• If the senior person refuses to revise it, let it stand for one cycle and revisit at grading with the data.

Note recurring patterns in your own coaching notes, not to call them out in the room. If a pattern surfaces three weeks running, address it privately with the lead rather than publicly in the check-in.

Homework for the team

None for this step. It is a lens, not a session. The team absorbs the vocabulary gradually by hearing the questions.

• Name the theme in your coaching notes: "This team defaults to activity KRs when under pressure."
• Design a single intervention rather than repeated in-session corrections: a short pre-read, a framing question at the start of Step 3, or a private conversation with the lead.
• Track whether the intervention shifts the pattern over three to four weeks.
• If it does not shift, escalate to a dedicated 30-minute conversation with the team rather than continuing to ask the question in passing.

• If the team disagrees that a pattern is present, do not press the point. Ask the one question once and move on.
• Note the Key Result and the disagreement in your coaching notes.
• At the end-of-cycle grading session, return to it: "This was the one we disagreed about. What does the data say now?"
• If the data supports the team's view, acknowledge it. If it supports the concern, ask what they would do differently next quarter.