OKRs: How Google Sets Goals
Vishal Sachar
Co-Founder & CEO of CLRT
Objectives and Key Results, the goal-setting method Google adopted early and made famous, is usually taught as a management ritual. The objective is where you want to go, the key results are the measurable markers that prove you got there, three or so per objective, each a hard number. That is accurate, and it misses the thing that makes OKRs matter more now than when Google adopted them. A goal you cannot measure is a goal you cannot delegate. In the age of agents, the key result is not a management ritual. It is the interface between your intent and the system that executes it.
Start with why most goal-setting fails. It fails at the key result, because people write key results they cannot fail. "Improve customer satisfaction" is a wish, not a key result, because there is no number that could prove you wrong, which means there is no number that could prove you right either. A real key result is falsifiable: a specific figure you either hit or miss, with no room to narrate your way to success. A goal you cannot fail is not a goal. It is a hope with a deadline.
Now connect that to where work is going. If you intend to hand execution to agents, and increasingly you do, you need a crisp, measurable definition of done, because an agent can no more pursue a vague goal than you can verify one. The same vagueness that makes a goal impossible to evaluate makes it impossible to delegate. A well-formed key result solves both problems at once: it tells the system what success is, and it tells you whether the system achieved it. Measurability stops being a nicety and becomes the thing that makes delegation possible at all.
An objective points. A key result proves. And the proof is what lets you hand the work to something other than yourself.
A deeper dive
A well-formed key result is structurally identical to an eval criterion, and noticing that is the on-ramp to the most important skill of the agentic era. Both are a measurable definition of success that a system can be checked against, which means writing good OKRs and writing good specifications are the same discipline (the argument in Evals Are the New Tests), and that discipline, specification quality, is becoming the founder's core competency. The classic OKR failure mode shows why: key results that measure activity ("ship five features") rather than outcome ("reduce churn to 4 percent") cannot tell you whether anything actually worked, and an activity is not an eval. The cascade matters too. Objectives flow down an organisation, but at each level the key results become the specification for the level below, and increasingly that level below is a set of agents, so a sloppy key result is now a sloppy brief handed to a system that will pursue it literally. The companies that win the next decade will not be the ones with the best agents. They will be the ones whose people can write the cleanest definition of done.
Work with CLRT
The scarce skill of the agentic era is writing a definition of done so clear that a system could execute and verify against it. That is the same skill as a great OKR, and CLRT helps leaders build it. Start with the goal that matters most to you this quarter.
Vishal Sachar
Vishal Sachar is the Co-Founder and CEO of CLRT, where he helps UAE businesses make sense of applied agentic AI and put it to work. He writes on agentic systems, AI governance, and the economics of automation. Reach him at vishal@clrtstudio.com or on LinkedIn.
