Foundations
Why OKRs exist, what separates an outcome from an output, and what a well-shaped OKR set looks like. These three ideas come up in almost every coaching conversation; getting them clear at the start saves a lot of rework later.
Why OKRs exist
OKRs are not a reporting format. They exist to do three things a team usually struggles to do on its own. Focus: commit to the few things that matter and say no to the rest. Transparency: make it visible what the team is going for and how far along it is, so nobody has to ask. Alignment: connect the team's goals to the wider direction and to each other. If a set of OKRs does not produce more focus, more transparency, and more alignment, the format is just overhead.
Open every engagement with Focus, Transparency, Alignment on the wall, then ask: "Is this giving us more of any of these three? If not, why are we doing it?"
- Write the three words visibly before anything else starts.
- When the team gets lost in mechanics, point back to the words.
- If none of the three answers "yes", pause the session and name that out loud.
Example: the same quarter, with and without focus
A team tracks 14 goals, all flagged as important, none moving. Nobody can name the top three from memory.
The same team commits to 2 goals for the quarter. Both move visibly each week and every team member can state them without checking a document.
Fourteen "important" goals is a queue, not a direction. Two that everyone can name is a commitment. The move from 14 to 2 is where focus actually happens.
Focus, transparency and alignment are the point. Every OKR mechanic is in service of them, and only them.
Outputs, outcomes, impact
An output is something you make or do: a feature, a workshop, a report. An outcome is a measurable change in someone's behaviour or situation that the output causes. An impact is the far-off goal that many outcomes add up to. Key Results live at the outcome layer. Outputs are too easy to claim as done while changing nothing; impact is too far away to steer against inside a quarter.
When a team proposes a goal, ask: "Is that a thing you make, a change in someone, or a far-off number?"
- Draw the three-layer ladder on the whiteboard: output / outcome / impact.
- Ask each person to place their proposed goal on the ladder out loud, before any discussion.
- They learn the distinction by sorting their own words, not by listening to a definition.
Example: an output that changed nothing
"We shipped the redesigned dashboard." Done, celebrated, closed.
"And did anyone use it differently after the redesign?" Silence. The outcome was never named, so the output proved nothing about whether anything changed.
Shipping is not the goal. The goal is someone doing something differently as a result. If you never name the change you expect in a person, you can ship indefinitely and still not know whether it is working.
Transparency is impossible if "done" means "we did the work" rather than "something changed for someone".
What a good OKR set looks like
At team level, the usual shape is one Objective and two to four Key Results, for one time-box, often a quarter. The Objective is a sentence anyone can repeat. The Key Results are measurable and few. Business-as-usual metrics such as uptime or support volume are tracked separately as health metrics, not crammed into the OKRs. Less is the discipline: a long OKR set is a focus failure wearing a planning costume.
Give the team a deliberately bloated example set and ask: "Which of these are really outcomes, and which belong elsewhere?"
- Print or write out one Objective with seven KRs, two of which are BAU metrics.
- Ask the team to cut it to the minimum set they could commit to and defend.
- Have them explain each cut out loud. Defending the removal teaches focus faster than any lecture on the ideal structure.
Example: from a wish-list to an OKR set
1 Objective, 7 Key Results. Three are tasks ("launch X", "migrate Y"), two track BAU metrics that never change much, one has no baseline. The set looks complete but gives the team no real direction.
1 Objective, 3 outcome Key Results. The tasks move to a project backlog. The BAU metrics move to a health-metric dashboard. The set is shorter but each item actually signals whether the Objective is moving.
A set of three that everyone defends is more useful than a set of seven that nobody can prioritise. The act of cutting is where the real alignment conversation happens.
Focus is visible in the count. If you cannot see the few that matter, neither can anyone else.