Most teams find out an OKR was weak three months too late, when the quarter ends and the score is meaningless. By then the objective has already shaped a quarter of work. The fix is to grade the OKR for quality before the quarter starts, not just for attainment after it ends. Those are two different acts. Attainment scoring asks how much you achieved. Quality grading asks whether the goal was worth writing at all, and you can do it in about two minutes per objective with a fixed rubric.
Here is the rubric I use. Five checks, each pass or fail. An OKR that fails any of them needs a rewrite before it goes on the board.
1. The objective is qualitative and memorable
A good objective is a sentence a person on the team can repeat from memory a month later. It says where you are trying to get to, in plain language, without a number in it. "Become the obvious choice for mid-market buyers" passes. "Increase qualified pipeline by 40 percent" fails, because that is a key result wearing an objective's clothes. If your objective already contains the metric, you have collapsed the two layers and left yourself no room to describe direction. Strip the number out and put it where it belongs.
2. Key results are measurable, with a baseline and a target
Every key result needs a number you can read today and a number you are aiming for. "Improve onboarding" is not gradable. "Lift week-one activation from 48 percent to 70 percent" is, because it carries both the baseline and the target. The baseline matters as much as the target. Without it you cannot tell whether 70 percent is a stretch or a rounding error, and at quarter-end you cannot say how far you actually moved. If a key result has a target but no starting point, send it back.
3. There are three to five key results
Two key results usually means the objective is thin or you have not thought about what "done" looks like from more than one angle. Six or more means you have listed tasks, not results, and diluted the objective until nothing is clearly the point. Three to five is the working range. It forces a real choice about what evidence would convince you the objective was met.
4. Key results are outcomes, not tasks
This is the check that fails most often. "Ship the new pricing page" is a task. It will be done or not done, and doing it tells you nothing about whether anything got better. The outcome version is "raise trial-to-paid conversion from 9 percent to 14 percent." The page might be how you get there, but the key result measures the result, not the activity. A quick test: if you could complete the key result and still have made zero difference to the business, it is a task. Rewrite it as the change you expect the task to produce.
5. Each OKR has a single owner
Shared ownership reads as accountability and behaves as its opposite. When two names sit on one objective, each assumes the other is watching it, and neither is. One owner per objective, named, with the authority to actually move it. The owner can pull in a dozen contributors, but there is one person who answers for the score. If you cannot name that person, the OKR is not ready.
Fixing weak key results
The pattern is almost always the same: a task pretending to be an outcome, or a target with no baseline. Two examples.
Weak: "Launch the referral program." Strong: "Drive 15 percent of new signups through referrals, up from 3 percent." The launch is a means. The percentage is the result you actually want, and now it is gradable.
Weak: "Improve customer satisfaction." Strong: "Raise NPS from 32 to 45." Same intent, but the second one tells you the starting point, the destination, and exactly how you will know in three months whether you got there.
Grade it before you commit to it
Run every objective through those five checks at the start of the quarter, not the end. If you want the checks applied for you, the browser OKR grader scores a pasted objective against this rubric in seconds, and the downloadable OKR grader skill runs the same rubric inside your own AI assistant so you can grade a whole set in one pass. When a key result fails the outcome test or is missing a baseline, the OKR writer skill helps you rewrite it into something measurable, and the OKR template gives you a structured place to keep the baseline, target, and owner side by side so the rubric stays visible all quarter.
A clean rubric catches the weak OKR on day one. What it cannot do is keep that OKR connected to the work once the quarter is underway, which is where most goals quietly drift from the projects meant to deliver them. Vindaris links each objective to the work that proves it, so a key result and its baseline stay tied to the real tasks moving it, and the quarter-end score reflects what actually happened rather than what someone typed in a hurry. Grade the goal well first, then keep it honest.