OKR Mistakes Founders Make and How to Diagnose Them
OKR mistakes founders make, sorted by what you see going wrong: a symptom-first guide to the four bad patterns and nine failures that turn goals to overhead.
The OKR mistakes founders make fall into two groups: four bad patterns in how key results are written, and nine failures in how the system is run. Every one is a choice, and every one can be reversed next quarter.
A plain list of mistakes is easy to nod along to and hard to use. This article is organised the other way round. It starts from what you can observe in your own company and works back to the likely cause, so you can find the mistake you are actually making instead of the one that sounds most familiar.
OKR mistakes founders make: where to look first
The guide's summary of all nine mistakes is the useful starting point. They are failures of discipline and connection, not of framework choice. That is good news, because it means no new methodology is needed. The remedy is fewer goals, written as outcomes, attached to the work and checked every week.
It also tells you how to diagnose. If the problem were the framework, you would look for a better one. Because the problem is discipline and connection, you look for the point where the goals and the work came apart. In the guide's account, that gap usually opens around the third week of a quarter.
Start with the symptom table below. Find the row that matches what you see, then read the section it points to.
| What you see | Likely cause | Where to read more |
|---|---|---|
| Everything was delivered and nothing changed | Tasks written as key results | How to write key results |
| Nobody can say whether the goal was ambitious | Targets with no baseline | How to set a key result baseline |
| Every goal is green, every week | Sandbagged targets or generous scoring | How to score OKRs |
| The goals are forgotten by week three | Goals kept away from the work | OKRs vs roadmap vs backlog |
| The list is long and nothing is ever declined | Having too many | How many OKRs a startup should have |
If delivery is perfect and nothing changed
This points to the defining error: writing tasks as key results. The guide describes work delivered in full while nothing changes as the exact outcome the method exists to prevent, and a task list wearing goal formatting brings it straight back.
Two of the four bad patterns live here. The disguised task is a key result that is a piece of work with a deadline attached. You can finish it completely and have nothing to show for it. The unownable key result is the cousin, one that no person or team can meaningfully influence, so it produces reporting rather than effort and teaches everyone that the goals are decorative.
To diagnose, take any key result and ask the guide's so-what test: if we achieved exactly this and nothing else, would the business be meaningfully different? If the honest answer is no, you have found the problem.
If the end of the quarter becomes an argument
When people disagree about whether a result was good, the cause is usually a missing starting number. Setting targets with no baseline means nobody can tell whether the goal was ambitious or already met, and the end of quarter conversation becomes an argument about interpretation.
This is the second bad pattern, the baseline-free metric, and the fix is cheap but must come first. Get the starting value, agree the source, and write the figure beside the target. If you do not have it, finding it is a legitimate key result.
If the board is always green
This symptom has several possible causes, and the diagnosis is a matter of asking which applies. The guide offers three for a uniformly green board: the goals were set inside what was going to happen anyway, people are managing the report rather than the work, or bad news is not travelling.
Behind those sit three of the nine mistakes. A sandbagged target is a number the team is confident of hitting. Scoring generously protects feelings for one afternoon and ruins the record. And tying goals to pay makes everyone rational about both, because hiding bad news and setting safe targets become sensible.
The sandbagged target is the fourth bad pattern, and the guide asks you to treat it as a signal to check your incentives rather than the team. Look first at whether the separation from pay and reviews is believed. The scoring side is in how to score OKRs.
If the goals disappear after the first few weeks
People work where the work is. A goals document that is not connected to the roadmap and the board is forgotten by week three, reliably. That is the mistake of leaving the goals somewhere the work is not, and the guide treats it as the mechanical detail that decides everything.
A close companion is skipping the weekly check. Quarterly goals reviewed quarterly cannot influence anything in between, which is where all the behaviour is. If there is no weekly rhythm, the guide says not to bother writing them. The rhythm itself is a fifteen-minute weekly check on the number, the owner's confidence and what is in the way.
If the structure has grown into a tree
Cascading mechanically means decomposing every objective into sub-objectives at every level until the organisation carries a bureaucracy of goals nobody reads. The sign is simple: people cannot say which goals are theirs, or the number of goals has quietly multiplied with the number of teams.
The guide's alternative is to align to the company objectives and not manufacture a tree. Each team picks what it serves and writes its own results.
If a goal everyone knows is failing is still on the board
This is never killing one mid-quarter. Carrying a dead goal to the end of the quarter because abandoning it feels like failure costs a full quarter of effort and teaches the team that the document matters more than the outcome.
The signal is easy to see in the weekly check: confidence has dropped and stayed dropped, and nobody has said the obvious thing. The remedy is an explicit permission to declare a goal dead, with the reason written down.
The nine, in one place
Here is the guide's full list of nine, grouped so you can see which ones travel together.
- Writing the goals badly: tasks as key results, targets with no baseline.
- Building too much: having too many, cascading mechanically.
- Setting the wrong incentives: tying them to compensation, scoring generously.
- Detaching them from the work: leaving the goals where the work is not, skipping the weekly check.
- Avoiding hard decisions: never killing a goal mid-quarter.
Notice that the groups share causes. If you find one mistake in a group, check for the others in it.
Reading the symptoms together
One symptom rarely appears alone. A board that is always green and a quarter in which nothing changed often come from the same set: goals written as work, scored generously, with targets nobody could test. The guide's grouping of the nine helps here, because the members of a group tend to travel together.
So when you find one cause, do not stop. Ask whether its neighbours are present as well. If tasks have been written as key results, check whether baselines exist, because a team that writes work as goals seldom writes starting values. If the set is too long, check whether it has been cascaded, since a tree is a quick way to produce too many goals. If scoring is generous, look at whether compensation is anywhere near the numbers.
The habit of looking for neighbours also stops you fixing the wrong thing. A founder who responds to an all-green board by demanding harder targets may be treating a symptom of fear that attainment is being judged. The target was set low for a reason, and the reason is the thing to change.
A diagnosis session in four steps
If you want to run this as a short exercise with your team, keep it plain.
- Write down what you saw last quarter in the language of the symptom table, not in the language of blame.
- Match each symptom to the likely cause and note which group of mistakes it belongs to.
- Test the cause against the relevant checklist line, and tick or fail it.
- Choose one fix for next quarter, and name who will check it.
Choose one, because the guide's closing point is that the remedy is small: fewer goals, written as outcomes, attached to the work and checked every week. Trying to repair everything at once repeats the original error of having too many.
Use the checklists to confirm the diagnosis
Once you suspect a cause, test it against the guide's checklists. The checklist for writing key results asks whether each one describes a changed state, has a baseline, has a single named owner who can influence it, and can be traced to roadmap or backlog items. The checklist for running the quarter asks whether the weekly check happened, whether anything dead was declared dead and whether scoring was done without adjustment.
Each line is a box you can either tick or not. If you cannot tick it, that is your next task. That is a more practical remedy than resolving to try harder.
Frequently asked questions
Which mistake should I fix first?
The defining one: tasks written as key results. If your key results are work, the rest of the system has nothing true to measure. Then fix the baseline, then the size of the set.
Are these mistakes permanent?
No. The guide says every one is a choice and every one is reversible next quarter.
Do I need a different framework?
The guide says these are failures of discipline and connection, not of framework choice, so no new methodology is required.
How do I know which mistake I am making?
Start from the symptom, as in the table above, and test it against the relevant checklist line. Most companies find that more than one applies, usually within the same group.
The full method, with the checklists, is in The OKR Playbook, a free 32-page playbook.