In a large, distributed organisation, OKRs rarely fail because people can't write a key result. They fail on history, structure, rhythm and fear. We've met every one of these more than once. Here are the fourteen that matter most, why they happen, and what helps.
One division runs OKRs, another runs KPIs, a region runs Hoshin and the business you acquired runs SMART. Each has defenders, and each defender is partly right. Rolled out from the centre, OKRs feel like a takeover.
What helpsConverge rather than install. Keep what each system does well and agree only what has to be common.
Converging goal frameworks →The teams that struggle most are the ones where the old system does the most jobs, woven into rituals, reporting and how managers manage. Teams newest to OKRs often go fastest, which can sting a little.
What helpsLead with the why, change by change, and show where the work that used to live in the old system will live instead.
How one group did it →Long-serving people have seen layoffs, restructures and change programmes that promised a lot and delivered disruption. They read every new system against that history, and others listen to them.
What helpsAcknowledge the history out loud. Keep the system visibly separate from reviews and headcount decisions, involve experienced people in shaping it, and retire the old practices that contradict it.
Strategy, operational improvement, discovery, keep-the-lights-on work and projects all end up in OKRs. Each is there for a sensible reason. Together they drown the signal.
What helpsGive every type of work its own home: strategic OKRs, operational excellence OKRs, initiatives, discovery and business as usual, each with its own reporting.
Enterprise OKR governance →A goal that needs three teams in different divisions, each measured on its own results, gets quietly deprioritised exactly when it matters.
What helpsCount the teams a goal needs before committing to it. One is ideal, two needs a single named owner, three or more is a redesign.
Count the teams a goal needs →Without someone deciding which goals get dedicated people, every goal gets a bit of everyone. Priorities are added and never removed, and resource fights go to whoever shouts loudest.
What helpsA portfolio cap, a priority order agreed in advance, and a rule that nothing new joins without naming what leaves.
Enterprise OKR governance →The same key result means different things in different markets and business models, so leadership spends its time translating instead of deciding.
What helpsA handful of shared measures, including at least one that shows whether you're winning, not just improving.
OKRs for multinationals →With AI in the work, many goals now take weeks, not quarters. Treat the quarter as the only moment to start or finish a goal and teams sit waiting, while moving people mid-quarter raids someone else's commitments.
What helpsKeep the quarterly rhythm, but make it a check-in and re-set rather than a big planning event. Let goals start and finish at the speed of the work, and when a function gives up people, cut its own goal in the same conversation.
The enterprise operating rhythm →Leaders are back to back. Planning gets squeezed into forty-minute gaps and produces incremental thinking, and distributed teams lose the corridor conversations small companies get for free.
What helpsTreat time for planning as a strategic problem. Measure diary congestion and context switching, run experiments to recover time, and protect planning from the top.
The enterprise operating rhythm →One good coach can't sit with thousands of people in several languages, so quality drifts by manager and the system slides into compliance.
What helpsClear quality standards, champions in every division, a learning track for everyone, and AI that coaches at the moment of writing.
Enterprise OKR training →Almost no meaningful outcome is moved by one team. Putting your name to a target that others also move feels like signing up to be blamed, so people retreat to outputs they control.
What helpsMake ownership mean stewardship, not sole cause. Pair outcomes with leading measures the team does control, grade with judgement, and have leaders go first with ambitious goals they talk about honestly.
When OKR scores are how a division is seen or rewarded, people write goals they know they'll hit. Ambition disappears quietly, and nobody announces it.
What helpsKeep scores out of reviews and bonuses, and align leadership incentives with strategic progress, not only revenue and margin.
OKRs and compensation →Much of what large organisations call strategy is a list of sensible operational improvements. OKRs then execute things that don't add up to a defensible position.
What helpsAgree what a strategy is before writing OKRs for it: the choices, and why a specific customer will choose you over a competitor.
The strategy choice cascade →Whatever you write at the start, you'll find a ritual that matters more than it looks, a team that needs a different starting point, or a meeting doing a job nobody named.
What helpsWork with people who know what great looks like and can adapt their way there. That only comes from having done it before.
Meet the team →Large organisations end up with safe, output-based OKRs not for lack of skill, but because people are protecting themselves in a system they have good reasons not to trust yet.
That's why the work is as much about trust and lived experience as it is about method. You can say the new way is safe, and you should write it down. People believe it once they've been through a good cycle.
Rarely because of the framework. They stall on legacy systems and habits, goals that depend on several divisions, missing portfolio governance, a quarterly rhythm that clashes with how people can move, fear of owning numbers others also influence, and scores being used to judge people.
Because established teams have the most to unlearn, long-serving people have seen change programmes before, and leaders have little time for planning. Adoption speeds up when each change is explained, scores are kept separate from performance judgements, and people experience a good cycle before being asked to believe in it.
Plan for around four quarters to converge a large organisation on one way of running, with value arriving well before the last team moves. The first cycle produces better goals; the rhythm and internal capability build over the following cycles.
Not directly. Wiring bonuses to OKR scores teaches people to set goals they know they will hit. Use OKRs to steer and learn, and design incentives around strategic progress and behaviours separately.
The OKR Implementation Review checks each one against how your organisation actually runs. It's free, and you decide what to do with it.