A Breath of Fresh Air: Project Oxygen and the British Army
It’s a simple question; an often-asked question that seems to have a million answers: what makes a good leader? It is worth asking this simple question, not because there is a simple answer but because asking it might just force you ponder a different and more important question: how could you ago about working out what good leadership is?
Answering the question starts with an assumption, one that we take for granted in the Army: the assumption that there are such things as good leaders, people who ‘do leadership’ better than everyone else. And yet we are told to challenge our assumptions. So what would happen if we try to disprove this assumption? If we set out to prove that good leadership does not exist and that good leaders do not matter when it comes to exceptional team performance?
This article is about just that – how one organisation asked “what if good leadership does not matter?”; “if it does matter, how do we measure it?”; and “if we can measure it, can we work out what it is?” The organisation in question is Google. The project that answered these questions was Project Oxygen.
Project Oxygen set out to prove that managers did not matter to performance but ended up proving that good managers were essential. It had a profound impact on Google and, even today, its methodology and findings are relevant to military leaders, how we develop leadership and to the Army Leadership Code.
‘…a breath of fresh air’
Project Oxygen began in 2008 when Michelle Donovan, a member of Google’s People and Innovation Lab, known as the PiLab, asked: “What if everyone at Google had an amazing manager?” The project’s name was taken from her early observation that “having a good manager is essential, like breathing. And if we make leaders better, it would be like a breath of fresh air.” Project Oxygen was born. And like every project in Google, it needed to deliver evidence based on a pretty high burden of proof.
Donovan and her co-researcher Neal Patel decided on a data-driven approach to work out what effect team leaders had at Google. First, they measured team leader performance reports. At Google, in comparison to the Army, these reports are heavily based on measurable performance metrics as opposed to subjective human ratings.
Then they measured how team leaders were rated by their subordinates. In Google this 360-degree-feedback-type measure was known as the team leaders’ ‘Googlegeist result’. The annual Googlegeist score measured how content Google members were with their job, but part of it focussed on the Googler’s assessment of their team leaders’ performance, conduct and support.
Under a sub-project called Project Gifted Youngsters Donovan and Patel examined the people who sustained the highest performance for long periods. First looking at the difference in behaviour between the top 4% and the remaining 96%, they later went deeper examining the top 0.5% against the remaining 99.5%.
Finally, they plotted team satisfaction (based on Googlegeist) against manager performance rating to identify those team with the standout leaders. Then they dug into which dozen-or-so behaviours on Googlegeist seemed to be the most important ‘leadership behaviours’ that drove team satisfaction.
Project Oxygen was getting pretty close to the answer. But the data still was not good enough to convince the engineering-minded Google team.
‘Even Google wouldn’t be crazy enough…’
Project Oxygen had worked out that teams with the best leaders were 5-18% happier than the average Googler, performed better and had a lower staff turnover. The problem was, this was just one set of data taken at one point in time. To genuinely measure the importance of leadership Google would have to shuffle all the teams up to see if those allegedly-great team leaders actually made a difference when their team changed.
As Laszlo Bock, Senior VP of People Operations later asked, “Even Google wouldn’t be crazy enough to randomly mix up teams and managers just for the sake of knowledge, would we?” Thankfully, there was no need. Just like another organisation you might think of, Google’s employees switch around between teams fairly regularly. It gave the Project Oxygen team the chance to test what kind of difference those leaders made.
And what a difference they made.
Sixty-five people moved to lower scoring leaders. Everyone one of them scored lower on their Googlegeist satisfaction score in the year following their move. At the same time, those moving from lower scoring to higher scoring leaders increased their Googlegeist satisfaction score. Most importantly, the satisfaction improvement was in areas related to retention, how they felt their career was developed and whether they trusted how performance was being managed in the team.
Having finally concluded (using Google-standard evidence) that good leadership mattered the project then decided to look at why these leaders were better. They did it in a most unGoogle-like fashion.
They just asked people.
‘a prescription for building great managers’
Google interviewed their team leaders, asking them what they did to increase performance and satisfaction. To keep the data pure Google used a double-blind interview system. Neither the interviewer or the leader being interviewed knew whether the leader was considered good, mediocre or bad.
The list of behaviours that fell out was then cross-checked against three other relevant data sets: Google’s annual ‘best manager’ awards, comments in the annual Googlegeist survey and a selection of peer-review feedback for team leaders.
To recap: Google discovered which leaders were both highest-performing and built the most satisfied teams. They measured these leaders’ attributes. They then checked a year later to see if the satisfaction of team members changed when they moved team. Having done this they asked team members what behaviours the best leaders exhibited, which they crossed-checked with three other data sets.
After crunching all this data Project Oxygen came up with eight attributes – you could call them leadership behaviours – that the very best team leaders in Google exhibited. It was, as Bock explained in his book Work Rules! “a prescription for building great managers”.
But let us stop for a second. They came up with a list? After all this data and analysis, after years of balancing and measuring, the best they could do is write a checklist? Surely this prescription for great leadership should have been more than that? This is another question worth answering, because Google is not the only organisation that provides its people with a leadership behaviour checklist.
The Checklist Manifesto
There is a lot to be said about checklists. We associate checklists with technical or medical applications, rather than human ones, but checklists come into their own when the task at hand becomes more complex than a person can handle without help. In his 2009 book The Checklist Manifesto Atul Gawande wrote about the introduction of the Model 299 long range bomber – an aircraft that would go on to give its users “a decisive air advantage in the second world war which enabled [the] devastating bombing campaign across Nazi Germany.” The only problem? It was so complex that the crews kept on crashing it. The US Army’s solution was a checklist. The checklist helped simple humans fly a complex aircraft and, guided by their list, pilots went on to fly the Model 299 – by then renamed the B-17 – over 1.8 million miles without a single accident.
We see the same thing in medicine. In a 2007 article Gawande talks about a 19-item hospital checklist that reduced surgical complications by 30% and fatalities by 50%. When procedures get too complex for a human to remember everything, we give experts a checklist to reduce accidents. And leadership is pretty complex. This was why Google chose a checklist. Just like the US Army in 1939, they wanted to reduce expensive training and replace it with a list of behaviours that, if followed, lead to success.
This is point worth revisiting. We might scoff at the idea of the British Army having a leadership code of seven behaviours but Google thinks a more-difficult-to-remember ‘leadership code’ of eight is worth having. In 2018 they even lengthened their checklist to ten behaviours.
(As an aside, Google also has a checklist of five behaviours a team leader should follow when a new Googler joins the team. They found that team leaders who followed the onboarding checklist helped their new team members reach full effectiveness 25% faster than those who did not.)
Equally, Google chose a behaviour checklist because it has another advantage. As David Garvin pointed out in his 2013 HBR article on Project Oxygen, “because the eight behaviours are rooted in action, it’s difficult for managers to fake them in pursuit of higher ratings.” It is hard to fake or cover up your actions.
The Ten Behaviours
Before explaining the ten behaviours (the original eight, and two 2018 additions) it is important to remember that Google has these ten behaviours because it used research to empirically prove that these behaviours improve performance, increase team satisfaction and reduce staff turn-over in their organisation. Google do not pretend these rules work for everyone but they know that they work for them.
However, human beings are pretty similar across the world. So, what can the British Army learn from these ten Project Oxygen behaviours?
1: Be a good coach
Interestingly, Google agrees with the Army Leadership Code booklet, which uses five of its 28 pages to explain the advantages of coaching. Coaching is not about teaching your people but helping your people learn faster and more effectively. It provides a compound advantage. So get yourself and your junior leaders on coaching courses. Google found that if a manager went on a coaching course their coaching score in their quarterly review went up by 13%. Or take another leaf out of Google’s book: it uses the top 10% of its leaders to coach the bottom 10%. If you find you have a great coach in your team get them to coach the other junior leaders. If they are that good, they should be able to handle the job.
2: Empower the team and do not micromanage
You just can not escape empowerment these days, yet Google was talking about empowerment back in 2008. Google defines micromanaging as ‘getting involved in details that should be handled at other levels’.
The question a leader needs to be able to answer is ‘what should be handled at other levels?’ There has been plenty written on the subject but some of the best has already been brought together in the article Know Your Role and Don’t be a NOBA. While the article discusses sub-unit command, it covers of the main point: Concentrate on adding the value only you can add.
3: Create an inclusive team environment, showing concern for success and well-being
Inclusivity is another phrase that can sound pretty on-trend. In 2008, Google’s third Project Oxygen attribute was ‘express concern for team members’ success and personal well-being’ but they updated it to include inclusion in 2018.
Showing concern for your people’s success and personal wellbeing fits nicely with the Army Leadership Model – also known as Adair’s Action Centred Leadership. In 1973 Adair wrote that being a successful leader was about more than achieving the task; a leader had to balance achieving the task against the meeting the needs of the individuals in the team. In 2008, Google proved it.
But even if you feel your boss cares about your well-being and success, you can not be happy if you are being excluded from the group. Humans feel a strong need to belong to a group. In the Army we build strong and cohesive teams that live and work together. Even more so than at Google, if you are creating a team that excludes some of its members you will not be getting the most from the team.
4: Be productive and results-oriented
But let us not be too fluffy here. Google also agrees with the top circle of Adair’s model. If you were not aware, Adair deliberate put the ‘task’ circle at the top of his model because it was the primary role of a leader. The best leaders are focussed on their teams’ output (what they produce or achieve) rather than the input (the number of hours they work).
5: Be a good communicator — listen and share information
Notice how Google does not just concentrate on communication as a top-to-bottom activity. For them, communication is about listening – to both subordinates and to peers. It is also about information sharing. In their 2014 book How Google Works Eric Schmidt and Jonathan Rosenberg describe a good leader as being “a damn good router” that takes information and gets it to the place where it is needed most. The give several rules for good communication.
The first is to ‘default to open’. Unless you are told otherwise, all information should be shared or sharable. Next is that the communications should be targeted at those that need the information, not just spread as widely as possible. Another is to relentlessly overcommunicate. Their guidelines for overcommunication include that the message should reinforce your core themes, should be authentic and truthful and should be fun and inspirational.
6: Support career development and discuss performance
Google discovered that its engineers hugely disliked being micromanaged when it came to engineering. On the other hand, they really valued being micromanaged when it came to their career development. People (not just software engineers) care about their career – as do you.
The thing is, most of us already try and help our team members to develop their careers. Helpfully, Google found two other things that are worth knowing.
First, they found that career development and performance discussions need to be different things if you want them to both work. Bock found that “combining the two kills learning”. Think about that next time you do an MPAR – an Army HR task that combines those two things.
Second, Google used the power in their team to provide career advice. They appointed Career Gurus who had specific expertise in a specific career area. These volunteers were the ‘go-to’ career advisors for anyone who wanted advice from outside the chain of command. Who are your unit’s career field experts? The officers who have worked in career fields, the SNCOs who have excelled in training establishments or can help explain to a JNCO the difference between aspiring to Brecon or aspiring to Sandhurst; these people are your unit’s Career Gurus. They can help you with the career development.
7: Have a clear vision/strategy for the team
While the Army is generally good at having a vision or strategy, it does not always happen at every level. Every leader should be able to answer the questions ‘why do we exist?’, ‘where are we going’ and ‘how are we going to get there’ (or ‘what are our principles for getting there?’) If you do not feel you can answer those questions, have a think about them. Your starting point is how your higher commander has answered those questions.
8: Have key technical skills to help advise the team
The Army Leadership Doctrine (via the AKX) makes this point very clearly. You do not need to be able to do the job of everyone beneath you but you do need to do your job and be able to advise those who work immediate below you. Never forget your role is not just to direct the people beneath you – it is to advise them as well.
(As an aside, how Bock describes this leadership behaviour is highly informative: “technical expertise was actually the least important of the eight [initial 2008] behaviours… Make no mistake, it is essential… But of the behaviours that differentiated the very best, technical input made the smallest of difference to teams.”)
9: Collaborate across Google
In his article on platoon command Chris Finbow wrote “…foster your own tribal ethos and identity but know that you are all part of a bigger one”. He is right but there is more to it than that. Your 1 Up wants you to succeed. But they also want to succeed as well, and for their 1 Up to succeed too. Google found that the most successful leaders collaborated across boundaries to add value to the wider team. Not only does cross-boundary collaboration help the wider company, it also builds cross-boundary relationships that help both collaborators gain value again in the future.
For an Army leader, this means building cross-boundary links at your level. It also means creating the environment where team members and subordinate leaders are encouraged and able to build cross-boundary relationships.
10: Be a strong decision maker
In Schmidt and Rosenberg’s How Google Works they explain what Google believes good decision making is made of. “[W]hen it comes to making decisions, you can’t just focus on making the right one. The process by which you reach the decision, the timing of when you reach it, and they way it is implemented are just as important as the decision itself”. As an organisation that makes decisions, gives orders, and expects things to happen, this is worth thinking about.
Do you consider those other factors when you need to make a decision? Do you start the decision-making process by wondering what process you need to make the decision, and when it needs to be made? And once you have made it, do you think hard about the implementation?
The ‘Google Leadership Code’
So, what can we learn from Project Oxygen’s data-driven leadership investigation? First, we can be reassured that much of what the Army teaches about leadership is broadly right: coach and develop your people and help them progress in their careers; care for them; build a cohesive and inclusive team; know how to do your job and how to advise your team; have a vision; demand high performance; communicate; make effective decisions; and do not micromanage.
But perhaps it is the second lesson that is the most informative. By using data and a little time you can genuinely come to a conclusion about which leadership behaviours deliver better outcomes in your organisation. Of course, one might conclude that Google’s leadership behaviours are effective across every group and in every situation. But Google know that they are definitely effective in Google. And that is what matters to them.
Leadership is effective when it is alive to the context in which it operates. By defining and measuring success metrics over time it is possible to come up with a short list – a checklist – of effective leadership behaviours that you can say, with confidence, work for your organisation.
At this point, it is easy to wonder about the Army Leader Code, a snappy checklist that happens to spell out the word LEADERS. Why could not the Army have carried out some research over a couple of years, measuring performance outcomes in an unbiased study to determine the behaviours that improved performance?
The answer is that they did. Along the way the eight leadership behaviours they discovered became seven, and changed slightly so that they could form the more memorable acronym, LEADERS.
Perhaps another article can tell the story behind the research that created the Army Leadership Code. How it started in the Commando Training Centre, was validated at the Infantry Training Centre and later rolled out across the Army Recruiting and Training Division.
Until then, Project Oxygen has offered every Army leader an alternative leadership code. Of course, it might be better suited to software engineers than Royal Engineers. But even so, the similarities suggest there is something there; in the way it was written, the way it was taught, and the way it finally proved that good leadership really does matter.