Seita

OKRs – Measure what matters

Is having goals and tracking performance good for your organisation?

Should you set goals for your organisation? Also for all departments and all the people in it? Should you even measure how everybody is doing towards those goals?

YES: You want to know where you’re going—it’s called having focus. And you want to know whether you are on course and are using sufficient acceleration. Maybe everybody in your organisation should be able to know.

NO: Having one or a few goals are good for direction, sure, but to track your progress on all levels and even for every person? That seems like a huge strategic decision to make. It also sounds like micro-management, which is not efficient and harmful for creativity and freedom.

I read “Measure what matters” by John Doerr, who used OKRs at Intel and brought them to Google early on. “Measure what matters” is not a handbook on OKRs, but more an introduction to a simple concept, plus a collection of case studies. It’s an illustrative read this way—but as there are not too many details, you’ll have to try out OKRs yourself. That is probably a good thing.

Enter OKRs

A model that can be useful for anybody on that YES-NO spectrum is getting popular recently: OKRs—Objectives and Key Results. I like this model because it offers a way to attain focus on day one, but there is no need to change the whole organisation at once, in order to test OKRs out. You decide how far you take it.

In the context of an organisation, the aim of OKRs is that anybody can answer the crucial question: “What is the most important thing to do?”. As a plus, they should also be able to argue why that task is the most important.

But let’s first understand the simple-at-heart concept: OKRs help you to increase focus by connecting a desirable objective to a few measurable key results. At the core, this is what an OKR is, and it already is useful! Here is a quick example to illustrate:

This example shows a natural language objective (“empower users”), and then goes into the countable key results, which are required to claim full success. Note that partial success is definitely in the picture, and that not all key results must be countable to get started.

In the OKR model, there are not many technicalities. Objectives are described as contextualised language (e.g. “Demonstrate that X is good.”), and they have a deadline, e.g. next quarter. Each objective is paired with a few key results, by which one can without a doubt measure how close one came, e.g. “write three benchmark tests” and “send twenty emails to key customers”. All OKRs are time-bound, usually by the same duration like a quarter (three months), so all OKRs can be evaluated together.

There should be only a few OKRs, which everybody can remember (at least the objective). At any time (e.g. bi-weekly), the state of the key results can be reviewed. For instance, a simple score can be calculated, to see which OKR is in danger of being missed. Google’s OKR scores are all between 0 and 1, for example.

When working on OKRs with customers, we at Seita have seen that simply formulating OKRs for the first time is already a very worthwhile exercise as it often makes you look at your ambitions in a new light. They are a great tool to pick up for a focus exercise – OKRs are therefore part of our Technology Roadmap workshop. From your first OKR session, which gives your team an initial wave of focus, the journey can continue to more refinement each quarter, lifting the focus of management up each time together with a deeper understanding of OKRs over time. We’ll update this example below.

OKRs in context

Let’s talk about how OKRs differ from two popular goal setting tools, where they came from and who might use them.

  • As opposed to S.M.A.R.T. goals, OKRs don’t require you to write a goal in a way that does everything at once. Those goals often become long, unreadable sentences. I appreciate how OKRs give you first the space to say where you want to be in natural language. Then, it’s time to roll up your sleeve and list all the things that constitute success in getting there, in a countable way.
  • As opposed to KPIs, Key Results have a measurable target, so you can ask for a progress update. By themselves, KPIs only show you a current state, without a definition of success or the share of available time already spent. Furthermore, OKRs tell you why you are measuring in addition to what to measure. With OKRs, everybody involved can agree faster, because they can see how measuring something is linked to a goal (ultimately, the overall company mission).

Doerr gives some context on how OKRs came to be. In the 1950s and 1960s, management guru Peter Drucker, who popularized the idea of management by objectives, transformed the way company goals should be treated. For instance, Drucker wrote that subordinates should be consulted on company goals, which was quite novel for the 1950s and 1960s. Then, Andy Grove at Intel developed OKRs and made the whole company use them in the 1970s and 1980s, to great success. Doerr then took OKRs to Google in 1999 when his company invested in them, and Google has been using them ever since.

As I said, OKRs are simple at heart, so one could start trying them out at any level, in any part of an organisation. One thing which makes them attractive in my opinion, is that they can also be combined with other useful management tools. For example, next to finding focus with OKRs, an organisation could easily make other choices, like having a Theory of Change to develop the overall objectives for the organisation, using Sociocracy to structure people and communication, and using Scrum within teams for agile knowledge workflow.

OKRs are embraced by commercial tech companies and startups, but recently the impact sector also became interested. For instance, Ann Mei’s book “Lean Impact” suggests using OKRs to all impact entrepreneurs.

Improving OKRs

I spoke a few times of the journey of using OKRs, where you can start simply and locally, and become better and spread them to other parts of the organization over time. Here, Doerr’s book is lacking some details. Let me illustrate with two angles what I’ve learned from my experience:

  • Binary versus countable: What comes most naturally to people is to identify the existence of something that wasn’t there before as a key result. For example, implementing a feature is binary and not countable, as either it is there (1) or not there (0). This places a high risk on that key result’s metric. Let’s say the feature has some minimal behaviour, but lacks a final touch. If you defined the key result in a binary form, the risk is that management would see progress here as 0%. Here is one idea: Feature completeness (for this evauaton period) can be measured by asking ten users to rate the feature at least 6 on a scale of 1 to 10. Another piece of advice for social entrpreneurs comes from Ann Mei: She stresses that when formulating key results, so-called vanity metrics (like “a hundred thousand people reached” or “two million dollars raised”) should not be used. Instead, use unit level targets like adoption rate of your service.
  • Action versus value: Also, listing activities comes most naturally to people practicing OKRs for the first time. And I believe that is fine. However, an argument can be made that what you really are striving for is to measure impact, and not activities. Advanced OKR practitioners therefore split key results (results from their activities which they cannot directly influence) from activities, which they agree to do (activities usually do not count towards the OKR’s score!).

With that in mind, let’s update our “spark joy” OKR example from above:

Here, we have increased the share of countable key results (versus binary) and only value-based key results remain. Directly influencable activities are listed apart.

Keep in mind, that there is no perfection here. Innovation is not completely predictable, you’ll also forget something anyway, etc. It is wise to not over-engineer OKRs (e.g. go for 100% countable key results). Try again next quarter, and keep improving! We’ll now focus on the management of OKRs across the whole organization, where you’ll find enough ground for more crucial optimisation.

Reasons to use OKRs

According to Doerr, OKRs have four “superpowers” (their names are his):

Focus and commitment
OKRs are a tool for focusing. By making the key results measurable, we stop having a binary note of success. We also get an idea of velocity. And by giving them a deadline they become a commitment. It’s pretty simple. Doerr emphasises that the leadership should start using them and never stop calling for their usage.

Align and connect for teamwork
OKRs can be used on all levels (company, departments, persons: CEO and employees). Their usage takes practice over time. If everyone eventually finds out how to align with the company goal, managerial or entrepreneurial thinking is trained. This also enables more engagement. Finally, an auto-detection of missing alignments is performed by everybody.

How to form goals across the organisation, so that they complement each other? Doerr shows a contrast: Cascading goals down the org chart (hierarchy) versus over-time conversion of goals, where bottom-up forming of OKRs should be enabled and cross-functional (horizontal) coordination is likely to happen. Doerr suggests maybe half top-down cascading, half bottom-up forming of OKRs.

Track for accountability
Transparency is important in OKR land. All OKRs are visible, so people will learn to only write down OKRs they can stand behind. This fosters early discussion of disagreements. It might require a mindset not everyone has, so in some case studies people tell how they decided to let some people go who didn’t believe OKRs worked for them.

Stretch for amazing
Are you into big scary goals? Then you can use OKRs to “stretch”. For example, Google looks at all OKRs in a traffic light visualisation. If every OKR is green, they say you’ve failed, because you’re not being ambitious. Google thinks reaching 70% of your OKR means you’re doing well. The book also contains case studies of entrepreneurs who like their OKR dashboard all green, though. I have something more to say on the effects of this goal stretching below.

How not to use OKRs

I’ve gone into some detail how to improve writing OKRs. On a higher (management) level, it is also important how to use OKRs across an entire organisation, so pitfalls from management practices of the 60s (when objectives initially increased productivity but soon felt too rigid and top-down) are not repeated with OKRs. Doerr cites a few wrong ways to use them:

  • Just using them as KPIs (no context)
  • Only top-down (no ownership)
  • Only tracked by accountants, not managers or employees (no real buy-in)
  • Tied to compensation or bonuses (reduces motivation)

This last one is the most crucial. Measuring goal performance has it uses, but not to judge how well a person is doing. Certainly not in a monetary way, but more crucial than that, not as primary tool in evaluation of any kind. Doerr cites Peter Drucker: “A manager’s first role is the personal one (…) the development of mutual confidence”.

In fact, Doerr spends a whole part of the book to argue that Human Resources (HR) and managers should use modern models like CFRs (Conversations, Feedback, Recognition) to engage with employees. OKRs and even CFRs should be decoupled from bonuses and compensation as far as possible (they do offer some information, but should not drive them), if the company wants to work effectively in the long term.

Some bad taste from the case studies

This book provides a Silicon Valley perspective, which for instance shows in its subtitle: “The simple idea that drives 10x growth”. Silicon Valley, of course, is obsessed with scaling fast. Substantially, OKRs are about organisation-wide focus and I do agree that scaling up is a time when you need focus. But I don’t believe that OKRs are fundamentally about scaling up organisations. OKRs are for everybody, certainly also for impact makers.

However, we should note that some of these case studies talk about practices that I don’t believe are in line with social entrepreneurship. These two examples come to mind:

Peer pressure at Intel
Andy Grove created an ambitious key result for sales targets in a time when Intel needed to beat its competitor Motorola. He also gave a bonus (trip to a tropical island) for reaching this key result, but any sales representative would only go if their whole team made the individual targets. This is for me a questionable management method (even if it is effective), and this part of the case study also was not essential to explaining OKRs in principle. It is not much more than Silicon Valley martial arts.

Watch time objective at Youtube
When Google started running Youtube, the Youtube team also had a stretch goal: one billion hours in daily watch time. This was a tenfold increase. They tweaked their algorithms relentlessly, so the site would recommend users videos to watch, to keep them hooked. All in the name of showing more ads. We know today what horrible consequences this had (Youtube radicalises its viewers as a direct effect). As I said above, I don’t see this as an OKR attribute, rather this is a capitalism attribute. In the end of the case study, they remark how they “may” soon pursue other metrics than watch time—maybe “user satisfaction” and “social responsibility”.