“Should I rollover OKRs which the team didn’t meet?”
No
“If I do rollover our OKRs, should I update them or keep them the same?”
Always update them
I recently spoke to an American health care provider about OKRs and backlog free working when these questions came up. This is not the first time I’ve been asked about OKR rollover so I thought I’d post my answer her.
Basically, when you start writing OKRs start with a blank sheet of paper.
A blank sheet has no carry over work. You are going to ask “What are the most important things we should be working on in the coming period?”
Do not start with carry over work, do not come with assumptions, do not make promises in advance. You have never set objectives with these people and this point in time before.
You probably set the previous OKRs 3 months ago. So let me point out the obvious: that world no longer exists.
Sure the last 3 months – January to April 2025 – have seen a lot of changes (understatement!) but every 3 months see change. Three months ago you probably agonised long and hard about what were your most important three (or maybe four) things that were worthy of OKRs.
Hopefully you have achieved these OKRs and have the outcomes to show for. But if not… do you really like in the same world you did 3 months ago?
Why would your priorities 3 months ago still hold today?
Is your business so stable that priorities from 13 weeks ago are unchanged?
Do your customers want the same things?
Are your investors and executive team relaxed about the market, cashflow, ROI and everything else?
Have your competitors, and government, really done nothing?
Count your blessings if you live in such a world. Lets pretend you do.
It is entirely possible that you are working towards some even bigger goal, something that does take more than 3 months to produce – landing someone on the moon, creating a new technology, gaining a certification/qualification, or some other big thing.
That is good, that is a bigger goal – the next level up, it might last a year, two years, 10 years or longer. That goal deserves articulating in its own right and while it remains your bigger goal all your quarterly goals should contribute towards it.
(Although I talk about the nested-egg and OKRs I haven’t posted about it here, I’ll correct that soon. In the meantime check out my Employees: growth driver or cost to trim post were I mention them.)
Now, suppose that the big thing you wanted 3 months ago is still really important to you. But, suppose you didn’t complete the OKR. There is work outstanding, or you didn’t meet all the acceptance criteria. Entirely possible.
Do you carry it over as is?
No.
Even if you didn’t complete it you probably did some work. If it is no longer important write it off, don’t fall for the sunk-cost-fallacy.
Assuming it is still important evaluate where you are and reflect it to reflect both the work done and what you have learned in the last 3 months.
You might be half way to your target already, in which case you might want to increase it. Or maybe you realise how hard the target is, in which case you might want to relax it.
Either way you will have learned something, revisit the OKR and ask: is it still worded the right way? are the targets right? should we add more key results or delete some?
So in short: do not, never, ever, look at an OKR and say “We missed it so cut-n-paste it for next time.”
In the early days of Agile this approach was called “Time boxing.” Work which wasn’t finished at the end of the time box was either released or abandoned. Over the years the industry has relaxed that understanding and today “time boxing” has been diluted to mean “a period of time.”
OK, I get that 2 week cycles are tough, I get that sometimes work flows beyond the sprint, I get that abandoning work is tough.
It is in working in time boxes, be they 2 weeks or 3 months, that we get good at working in time boxes. They make our lives difficult to force us to think, revaluate and learn. We get better at it the more we do it.
So please, lets not unbox OKRs: 3 months is a long time. Just ask any politician.