4 key lessons I’ve learned about OKRs
At Elsevier, we want everyone in the company to be able to describe how their work contributes to our overall company goals and how it aligns with the work of their peers.
Objectives and Key Results (OKRs) are a methodology for setting transparent, measurable goals that helps ensure that people across an organisation focus on the same priorities and track their progress towards those goals.
- Objectives are what you want to achieve.
- Key results are how you will know if you’ve achieved your objective, the measures you track.
More than 100 teams in Elsevier already use OKRs, and that number is growing. So they’re not a new thing for the organisation, but I only joined a few months ago, and they’ve been new for me. And, as someone who works with organisations setting objectives and measuring performance, I’ve enjoyed learning about them.
This is a post to share the 4 key things I’ve learnt about OKRs— my top 4 take aways. These are my key personal learning points — they’re not meant to be a comprehensive list of what you need to know in order to implement OKRs.
There’s plenty of other important stuff to understand about OKRs and to think about if you’re going to use OKRs: how to write a good objective, how to score KRs, how to use confidence measures, how to adopt an effective timeline/rhythm/cadence, how to track OKRs and structure check-ins, etc. If you want to learn more about those things, you can take a look at my reading list.
Lesson 1: Keep activities and results in separate buckets
OKRs focus on the outcomes you want to achieve and not on the things you’re going to do to achieve those outcomes.
It was Felipe Castro who really helped me to understand the importance of this distinction by talking about buckets. He explained that organisations need to manage two different “buckets”. The first bucket is for your OKRs (the outcomes you want to achieve and what you’ll measure to know whether you’ve achieved them). The second buckets is for the things you are going to do to try to achieve those outcomes — the activities.
For years, I’ve managed performance reviews for organisations, and these have covered what we’re trying to achieve (objectives), metrics (key results), and initiatives (activities such as projects and programmes). But often I’ve had it back-to-front, starting with what we’re doing in a particular area and then seeing if it’s being successful/on track to be successful.
By starting with what outcomes you’re trying to achieve, you focus first on results. And that means that you judge your activities in a different way. You’re much less wedded to the activities you’re currently doing (eg your current portfolio of programmes and projects), and you’re freer to ask whether they are the right ones. Activities thereby become much more like experiments — and every time you do something you’re regularly checking to see whether it’s having an impact on your key results. If it does, you can double down on it. If it doesn’t, you can quickly stop it and change course.
Lesson 2. Don’t cascade OKRs: Engage people and ask teams to align
I’ve been involved in many exercises to cascade objectives. You start by setting the top-level objectives, and then you cascade them down the organisation, one level at a time. Cascading requires objectives at every level in your hierarchy, and you can’t set the objectives for your level until the level above you has set and published theirs.
I’ve experienced 3 main problems with this process of cascading objectives:
- It’s slow: it can and often does take months.
- It’s opaque: people see the objectives of the level above them, but often they can’t see the objectives further up the organisation, or the objectives for teams in a different part of the organisation. Instead of being a unifying methodology that helps with goal alignment, this lack of transparency can breed suspicion (“what are the CEO’s objectives?” “what are the priorities for that team and do they conflict with my team’s objectives?”), and it can be hard for a team to see how they contribute to the organisation’s goals.
- It reflects a top-down, command and control way of working, which can be dis-engaging when you want to empower more autonomous teams.
In an OKR framework, however, the top-level goals are clear and transparent, and each team can align their OKRs directly to the company OKRs, skipping all the layers of hierarchy in between. Any team can also see the OKRs of any other team.
- How can we best contribute to the organisation’s annual OKRs?
- Which of the Key Results included in the annual OKRs can we inﬂuence?
The result is a faster, less bureaucratic process that also increases engagement. It makes it easier for teams to see how they contribute to the organisation’s goals and what others are focusing on. And it means that people align around outcomes, not their org structure.
As Laszlo Bock, Google VP of People Operations, put it:
“Having goals improves performance. Spending hours cascading goals up and down the company, however, does not.”
Lesson 3. Keep OKRs and your operational metrics separate
OKRs can become the flavour of the month, and some people want to turn everything into an OKR. But not everything has to be an OKR! As Christina Wodtke explains:
“OKRs are just a framework for creating and ensuring focus on what really matters, but not if you stuff them full of every single business-as-usual initiative you have going. OKRs are not a way to control the way your employees spend their time; it is a way to share your vision so they can make their own judgement calls.”
In particular, keep your OKRs distinct from your BaU operational metrics.
Operational issues (eg keeping the lights on, or meeting an SLA) are important for an organisation, and you need to report on and monitor operational metrics. But they don’t need to be expressed as an OKR.
For some areas of operations, you need to operate at 100% and that’s what you’re achieving, in other areas you might be operating at 50% and that is sufficient and you don’t need or want to improve performance in that area. Not everything needs to be pushed. Some things need to be maintained. As Christina Wodtke puts it in her book:
“OKRs are not the only thing you do. Trust people to keep the ship running, and don’t jam every task into your OKRs.”
I’d add to this: “and don’t jam every metric into your OKRs”. You will continue to monitor your operational metrics. And, as long as they stay within whatever threshold you set, you don’t have to do anything different about them. (This distinction should not mean that operational metrics are second-class citizens to OKRs.)
There is a caveat to this distinction/separation between steady state/incremental improvement BaU metrics and OKRs. If you want to transform how you’re doing operation X and/or achieve a step change in operational performance in area Y, you can take an operational metric and turn it into an OKR.
When setting your OKRs, you can look at all your metrics/KPIs and choose the ones you want to take to the next level of performance. Those are your OKRs.
Your OKRs are above and beyond your BaU — stretch objectives directly connected to your company’s goals. Failing to differentiate between commitments that you have to meet and stretch OKRs is the #1 trap in Google’s OKR playbook.
And remember: not everything can be a stretch at the same time. Setting too many OKRs is the #1 common mistake that Christina Wodtke has seen in companying implementing OKRs. A key purpose of OKRs is to focus on a small number of your most important goals (“less is more”, as John Doerr puts it). And I guess we all know what it feels like to be stretched in too many directions at the same time…
Lesson 4: Keep OKRs and compensation separate, but not completely separate
It is a central tenet of the OKR framework is that OKRs are separate from compensation. This is a mantra repeated by many of the key writers on OKRs, from Andy Grove to John Doerr, from Christina Wodtke to Felipe Castro.
Typical reasons given for this separation are:
- people may try to ‘game the system’ (eg setting OKRs that easy to meet within a team’s own existing headcount and budget, rather than truly stretching),
- people will be less likely to take risks and try new things, reducing experimentation, and
- people may focus on OKRs that are fully under their control as opposed to ones that depend on others.
I got all of those reasons. But nonetheless, it took me a long time to get my head around the separation.
This was because of one nagging question that I kept coming back to:
If you’ve got your OKRs, and you’ve also got separate objectives for your annual performance review/compensation, aren’t we back with two of the key things OKRs are often implemented to avoid: a lack of transparency, and multiple sets of (possibly conflicting) objectives?
I got past this when I understood two things:
- OKRs and individual compensation are not completely separate, and
- your performance review/compensation isn’t based on a different set of objectives, separate from your OKRs and pulling you in a different direction.
Rather, your OKRs are one input used to determine how well an individual is doing. (Other inputs might be behaviours — how someone goes about their work — and 360 degree feedback.) That is, how you perform against your OKRs can influence your performance rating and any compensation review, but not determine it. So they’re one factor out of several.
Two further reflections:
(i) How new & different really are OKRs?
Some people paint OKRs as a radically new tool, like they’re an idea that is radically different from what’s gone before. It’s easy to say that the old way of doing something (in this case, management by objectives) is bad, and the new way is different and good:
But in this case that’s really misrepresenting what’s gone before and just setting up a straw person to knock down.
It’s not new for objectives to be tied to measures.
It’s not new for objectives to be quarterly. At the Food Standards Agency — a regulator and a small government department that is part of the UK civil service, hardly a silicon valley start-up or a place where you’d automatically look for the latest cutting-edge management thinking — we’d review objectives and performance measures quarterly.
I’m not denying there are differences between traditional management by objectives and OKRs. But management by objectives done well can be better than OKRs done badly. Which leads me on to…
(ii) OKRs are not a silver bullet
Some people can come across as rather evangelical in their enthusiasm for OKRs. They suggest that introducing OKRs will transform your organisation. It’s as if they’re saying: Implement OKRs and all your problems will be solved!
People share case study after case study of organisations that have found success after adopting OKRs: They were behind the success of Intel! They made Google successful! Twitter uses them! Facebook uses them! Bill Gates “wields OKRs to fight devastating diseases”! Bono “deploys OKRs to save lives in Africa”!
This feels like a variation on trait theory — you look for the common trait/management practice in a group of successful companies and extrapolate from that if you adopt that management practice then you too can be successful…
The reality of course is somewhat more complex…
And you hear rather less about the companies that used OKRs but failed.
John Doerr recognises in his TED talk that “OKRs are not a silver bullet.” And in his book he emphasises that they can be executed “well or badly”. Indeed, he dedicates several chapters to culture and continual performance management, his key point being that OKRs “cannot substitute for sound judgment, for strong leadership, or a creative workplace culture.”
No management methodology in the world is ever going to transform your organisation if those other things are not also in place.
When it comes down to it, OKRs are a tool to have in your management toolbox, albeit they have the potential to be a very powerful tool.