OKRs: Common Mistakes and How to Avoid Them
Objectives and Key Results. Simple enough in name and theory - but what do OKRs mean in practice? And do organisations really need to use them?
Contrary to common belief, OKRs are not measures of performance, although they are often used this way.
OKRs assist organisations in setting a small number of change goals (the O), and ways of determining progress to those goals (the KR); and are used to make the big objectives that need cross-functional collaboration to succeed more clear.
Andrew Ormerod, Tech Nation’s Specialist Team Operations Coach says if OKRs are used strictly in this way, they can be very effective. But more often than not, OKRs aren’t used in the way they’re supposed to be - and that’s where the problems start.
The good and the bad
Whether you’re already using OKRs within your company, or are thinking about incorporating them in the future, there are a few things to think about.
What OKRs can be good for:
- Setting big goals for a time boxed sprint (which doesn’t have to be 3 months!)
- Having ways of measuring progress of those goals
- Getting smaller units to think about how they contribute to the big goals - what is their part? How do others depend on them? Who else do they need to work with to make things happen?
Some of the potential pitfalls of using OKRs:
- Using "KRs" as measures of performance - or even worse - tying compensation to them
- Cascading objectives so each team can now look only at its own OKRs, without reference to the global ones or to other teams
- Working to the KR and losing sight of real world outcomes/customer value that you’re trying to achieve
- Tying every part of the business to the same quarterly schedule
- Having no 'second loop of learning' to regularly check if you still have the right goals
- Losing focus on opportunities emerging outside pre-defined goals
- Using any kind of framework or metric as a way to manage individual or team performance without going to see for yourself as a manager what is going on in their work
OKRs are not performance measures
OKRs aren’t for measuring performance or managing business as usual (BAU), Andrew notes.
When it comes to performance, Andrew says you should think about measuring customer value and system performance (e.g. end-to-end time).
All measures are distorting to a work system, and these measures are the least distorting, since they are closest to the true purpose of a system: to deliver as much value to customers as quickly as possible.
How many customer issues are you solving at the first point of contact? How quickly are you putting a product in the hands of your customers after they order it?
The DORA metrics, for example, are spoken highly of by Chief Technology Officers (CTOs) who think about measuring performance of development in this way.
They look at how much failure demand you create when committing code - how much time are you spending fixing problems that you created with previous work vs how much of your time are you creating new value for customers? How easy is it for you to fix things when they go wrong?
Andrew says: “Your goal should be to track performance numbers over short cycles (weekly, daily), plot them as trends, and reduce variation before seeking to drive improvement. It’s not enough to say ‘the number went up this week compared to last week’.
“Any system that is made up of multiple sub-systems will exhibit normal variation. The number will be up some weeks, and down other weeks. But within what range does it vary? How predictable is it overall?“
“If you use ‘key results’ as performance measures, then you have lost the ability to see whether you are reducing variance in performance (reduced variance = greater reliability = more ability to plan) - because your KR progress can only go up linearly across a quarter, before resetting at the next quarter.
“You also lose sight of the fact that what matters is creating customer value: people start working to the measure, regardless of the real world meaning that this has.
“In short, not meeting your KR is not the same as a system performing badly, and meeting it is not the same as a system performing well! (Particularly if you stick to the rule of setting ambitious targets that you are not going to hit every time).”
Quarterly objectives: Why you need to be more flexible
If you ‘cascade’ objectives, where each department takes the big organisational goal at the beginning of a quarter and sets its own goals, your ability to drive cross-functional collaboration across the organisation will diminish.
If each department focuses on its own goal, without referencing other departments, you won’t solve the problem of siloed work.
Andrew says: “You cannot use half a bridge. If one department meets its objectives, and another doesn't, and the overall objective relies on both parts coming together...then all you've done is created a lot of work in progress.
“Having too much 'work in progress' (for example, incomplete inventory - stuff that cannot yet be shipped to customers) is how organisations go bankrupt. This is just as true in knowledge work as it is for physical hardware.”
Although the working world breaks the year into four segments, Andrew notes that quarters are arbitrary periods of time, because let’s be honest, the real world and its events are not neatly distributed into quarters.
He says: “Things change all the time - and different departments have different rhythms that arise out of their work. Are there parts of the business that would be better revisiting goals on an annual basis? Are there others where change is rapid and a monthly cycle would work better?
“Sprints are a good source of discipline (working in time boxes - committing to push the needle on some subset of outcomes for a period of time rather than jumping around all the time), but aligning everything rigidly to the same calendar schedule is not a way to build a responsive organisation.
“It’s better to have big organisational goals in longer sprints, and then shorter cycles as you get closer to the shop floor. And you should always have some mechanism for checking, ‘Is this goal still the right goal for us to be moving towards?’”.
It’s hard to set good goals, define good key results, and good performance measures Andrew says.
“It is time-consuming stuff - and involves not falling into one of several traps discussed above. If your organisation is committed to setting OKRs, the cost for managers not doing it is immediate and visible. If they don’t, they will likely get in trouble. So they do it quickly, in order to get it done - without having time to think about whether they are doing it well.
“But the cost of doing it badly comes a long way down the line, and can be much much higher in terms of distortion to the quality of the work, loss of morale, etc. That delay makes those costs harder to attribute to their true cause: bad OKRs.”
Andrew adds that managers and leaders alike should beware of frameworks that stand in for doing the real, time-consuming work of managing teams and collaborating; spending time with your reports and colleagues; seeing what is going on for yourself.
He says: “Thinking ‘if we just have an app, then we can see it all in one place’ meaning ‘we never have to go down to the factory floor, we can just manage from the executive office,’ is an error.”
Objectives and Key Results are only a goal, Andrew concludes.
“They can help teams and individuals to focus, but they are not actual outcomes, yet most organisations get caught up in the act of having them.
“They spend too much time and effort planning to meet them, and in the process, they delay delivery and run out of time and money to complete the feedback loop."