How OKR works in practice

In the previous article we covered the history behind OKR: Andy Grove at Intel, John Doerr carrying the model forward to Google. This article is about the mechanics itself — how to actually write an OKR that does the job it's meant to do.

Objective: the direction

Objective answers the question what. It should be qualitative, not a number — more a direction that's clear enough to understand and ambitious enough to feel like a real goal. A good Objective should inspire, not just inform.

"We will reach profitability this year" is an Objective. It says nothing about how much, but it says clearly where we're heading.

Key Result: the proof

Key Result answers the question how do we know we succeeded. Here it needs to be measurable, concrete, and time-bound. If the Objective is the compass, the Key Result is the odometer.

Tied to the example above, the Key Results might be: margin above 20 percent, customer churn below 5 percent, product cost reduced by 5 percent. Three numbers that together prove the direction held.

The rules for good Key Results are usually summed up like this: specific, time-bound, ambitious yet realistic — and measurable. Otherwise it isn't a Key Result, just a wish.

The difference from ordinary goal models

Classic goals are often set annually, top-down, private or per department, and directly tied to bonuses. OKRs are more often set quarterly or monthly, are open for the whole organization to see, are built bottom-up or sideways between teams — and are deliberately kept separate from bonuses. The point is to create a higher tolerance for risk: if an Objective is genuinely ambitious, it should be okay not to reach it fully — otherwise you're only setting goals you already know you can hit.

How to actually use it

A few practical rules of thumb tend to work better than others:

• Max three Objectives per team or person at a time. More than that and focus disappears.
• Max five Key Results per Objective — three tends to be the optimal number.
• Set them quarterly as the base rhythm, shorter when change is fast.
• Review weekly for quarterly goals, monthly for annual goals.
• Start small: one single Objective until the model settles in, rather than filling the whole organization with OKRs at once.

Objectives can also be set at several levels at once — an aspirational vision spanning several years, strategic OKRs annually, and tactical OKRs quarterly that break down into weekly initiatives. The further down that chain, the more concrete the goals become.

Three ways to set Objectives — and why it matters

How an Objective gets set affects how much it actually engages people:

• Top-down: the manager decides, the employee executes. Fastest to implement, hardest to generate real engagement around — especially if it feels like an order rather than a shared goal.
• Negotiation: manager and employee meet and shape the Objective together. A middle ground that often balances direction with participation.
• Bottom-up: the employee identifies what needs to be done based on what's already known about the organization's goals. Generates the highest engagement, but requires the strategy to be communicated clearly enough that everyone can draw the right conclusions on their own.

Pure top-down control tends to produce OKRs that are, in practice, just renamed task lists. The point of the model — engagement, ownership, a real sense of working toward something — gets lost if every Key Result is just an instruction in new packaging.

Prioritizing with traffic lights

A simple way to track several Objectives at once is to color-code them like traffic lights: green when on track, yellow when uncertain, red when it's time for the whole team to pitch in. Red isn't a failure in itself — it's a signal of where attention is needed to get movement back toward green.

A handbook for implementation

Where Doerr's "Measure What Matters" is narrative and inspiring, "Objectives and Key Results" by Paul Niven and Ben Lamorte is more of a handbook. It digs deeper into how to actually introduce OKR in an organization, what the most common implementation pitfalls are, and how to avoid the model turning into just another reporting burden. A good second step once the basic mechanics have settled.

Next: OKR vs KPI

So far, OKR might sound like just another way of measuring things. But that's where it gets interesting — because OKR and KPI actually measure different kinds of reality. One tells you what already happened. The other drives you toward something that hasn't happened yet. That's the subject of the next article: OKR vs KPI.