Objectives and Key Results

For a long time companies wrote goals down and then nobody opened the document again until the year was over. A line like "we grow the product this year" would go up at the January all-hands, and by December there was no way to answer whether it had happened. Nothing in that sentence could be checked.

Year-end reviews had a predictable shape because of it. Marketing said visitors had doubled, engineering said twenty-six features had shipped, sales said contracts had come in under plan. Each team had brought the number it looked best on, so nobody could conclude whether the year had gone well. Which number to watch had never been settled back in January.

Objectives and Key Results (OKR) removes that by splitting the goal into two fields. One field holds the direction, the other holds the check. The objective says in words where the team is going. The key results attach numbers that settle whether it actually went there. After the split, the December question becomes a matter of looking rather than a matter of interpretation.

Openness was in the design from the start. When what every team wrote is visible to everyone, two teams are less likely to do the same work separately, and a team is less likely to sit waiting on something another team has quietly deprioritized.

Settle the objective two or three weeks before the quarter starts, attach key results to it, then keep reopening that same document through the quarter to update where things stand. There are five steps.

1. Pick the objective. Write it in words rather than numbers, and keep any one group under three for the same quarter. Pagedeck's objective last quarter was a single line about getting teams on the free plan to start paying.
2. Attach two or three key results. Something like moving paid conversion from 4 to 7 percent. Write the change your work produced, not the work itself. "Onboarding redesign shipped" doesn't qualify, because the redesign can land and conversion can stay exactly where it was.
3. Settle the company-level items first, then have each team write in its own words which part it is taking. That holds up better than slicing the company goal and handing pieces down. People reread sentences they wrote and skip sentences they were handed.
4. Update progress once a week, in about fifteen minutes. What gets looked at is whether the number moved, not how much work got done. If it hasn't moved, change the approach.
5. At the end of the quarter, score each item between 0 and 1.0 and write a paragraph on why it landed there. The scores feed the choice of next quarter's items, and the items themselves usually get replaced.

Agree on the height before anyone writes. If the team is aiming at a stretch goal, the agreement that 70 percent still counts as a good outcome has to exist before the objective is written. Without it, people write the line they are sure of clearing.

It fits organizations whose priorities genuinely change every quarter. When an annual plan goes stale in three months, a structure that gets rewritten quarterly keeps up with that pace. It also helps when there are enough teams that people don't know what the others are doing — one open document replaces several meetings.

The places it doesn't fit are equally clear.

1. A number that can't move inside a quarter makes a poor key result. Brand awareness won't shift measurably in three months, so forcing a number onto it pushes effort toward changing how the measurement is taken.
2. Repetitive operational work is a bad match. Answering customer questions within a few hours is a line to hold rather than a goal to raise this quarter, and values like that belong on a standing indicator.
3. Under about ten people, the overhead outweighs the return. At a size where everyone already knows what everyone else is doing, writing the document costs more time than it saves in coordination.

The most common failure is wiring this straight into evaluation and pay. Once attainment affects salary, people write only the lines they can clear, and the height this method was built to encourage disappears. So teams share the material with performance reviews without tying the two together in a formula.

"That's a task, not a key result. Rewrite it as the number that moves when the task is done."

Putting completed work in the results field leaves nothing to judge at the end of the quarter.

"Three objectives is too many this quarter. Can we drop one and push it to the next?"

How many items you allow yourself is the prioritization — a list that includes everything has none.

"0.7 is a good outcome. If we keep landing on 1.0 we should be aiming higher."

Reading the score as a signal about the height of the goal rather than as a report card.

Pagedeck's first quarter with this was close to a failure. Grace wrote three company objectives and each team attached items underneath, and by the second month almost nobody had opened the document. No weekly update had been scheduled, and only half the company knew where the file lived.

Two things changed in the second quarter. The objectives went from three to one, and fifteen minutes in front of the Monday all-hands got dedicated to updating them. In those fifteen minutes Mia read out conversion, Ravi read out traffic, and Sofia read out signed contracts, and it became visible that the three numbers were different stretches of the same goal. The quarter scored 0.65, and that visibility is why the team counted it as a success.

Filling the key results with a task list is the most common mistake. Tasks are easy to write and easy to mark finished, so that is where the hand goes. But write three lines like "payment module improvements shipped" and the quarter ends with all three complete and paid conversion still at 4 percent. The document survives; what changed because of it is something nobody can explain.

Reading 100 percent attainment as a good outcome is the other one. Goals exist to be met, so the reading seems obvious. But this method is usually set up so that 70 percent counts as strong, and the height of each goal is chosen on that assumption. When every team keeps landing 1.0, the goals were set low rather than met well. Leave it alone and within two or three quarters the process is just transcribing work that was already going to happen.