How do modern companies measure performance? To answer that question, I turned to Measure What Matters, the book widely considered to be the authoritative source on OKRs (Objectives and Key Results), the tech industry’s most prevalent goal-setting framework. OKRs have been widely adopted as a way to define goals and measure success towards them, but what many people may not realize is that the framework also involves a set of techniques to manage and motivate employees. Below is a summary of the book’s key ideas in TL;DR format.

Setting Goals

Under the OKR Framework, the top-level goals that we’re aiming to achieve are called objectives. When setting our objectives, the general guideline is to limit the number of objectives that we pursue at any given time to between three and five. The purpose of this upper limit is to make us scrutinize our priorities to identify the most important, high-leverage ways that we can spend our time. If you’re dividing your time between ten different things at once, then chances are that less-important things are probably distracting you from more important things. Conversely, if you clearly define the few goals that are actually of the utmost importance to you, then that can help you to more effectively concentrate your attention and get into a psychological state of flow with your work.

My opinion on the “3-5 objectives at a time” guideline is that it makes intuitive sense, but requires tact and diplomacy to put into practice in a real job setting. For example, if a co-worker were to ask you to collaborate on a project that wasn’t tied to one of your few committed objectives, you would have to figure out a polite way to decline that request.

Measuring Progress Towards Goals with Key Results

An objective give us the destination that we want to reach, but not the path to get there. To fill that gap, we need to come up with a set of Key Results for each objective. Key results are the set of concrete, measurable outcomes that we need to make happen in order to achieve the objective. Key Results serve two primary purposes. The first purpose is to create alignment by ensuring that everyone agrees on a particular approach towards achieving each objective. The second purpose is to give us a yardstick against which we can track our progress.

Here’s an example of how one might pair key results with an objective: let’s say that our objective is to make our website load faster. The associated key results could be: 1) implement a skeleton loader that displays within 2 seconds of page load on a 50MB/s internet connection, 2) minify the page’s CSS down to 3kb, and 3) compress all images down to 75kb maximum. The objective of making the website load faster is very general, and it could be achieved through different approaches. The key results function to align everyone to a specific approach, and provide a basis for measuring how well the objective is being met.

One word of caution about key results: they should always be written with guardrails in order to prevent misaligned incentives. For example, consider the key result of “build one new feature for the app every month”. The fact that the metric mentions speed of development, but not quality of software, could inadvertently create an incentive for developers to compromise quality for the sake of speed. To disincentivize that type of behavior, the key result would be more effective if it were written with a guardrail such as “build one new feature every month with zero critical defects”.

Creating a Culture of Achievement

Now that we’ve covered the nuts and bolts of what an OKR is, let’s move on to the author’s recommendations about how OKRs can be used to create a broader culture of achievement.

From the start of the goal-setting process, people should always have some say in defining their own goals, as opposed to passively receiving a set of already-defined goals from their boss. Allowing people to define their own goals serves two purposes. First, it serves a psychological purpose, because people are more motivated to work towards goals that they set for themselves than those that they are assigned by authority figures. Second, it harnesses the collective intelligence of everyone in the company to set smarter goals. The person doing the actual hands-on work might have a better grasp of the ins and outs of the relevant subject matter than their manager, and that could allow them to look at things from a different angle and perhaps suggest better ways to work towards a goal. The door to that possibility is opened when you allow employees to collaborate in defining their own objectives.

Goals should also be created in a way that pushes people to fulfill their full potential. To that end, the book recommends including some “stretch goals”, which are goals so ambitious that they are unlikely to be completely met. Stretch goals have played an indispensable role in triggering the most groundbreaking initiatives in the history of science and technology. Transformative technological leaps can only be made if someone sets out to build something radically superior to the status quo, which necessarily implies a high risk of failure. Even though the risk of failure can seem daunting, there’s a large body of psychological research that seems to suggest that challenging goals cause people to perform better and feel more engaged, even if they aren’t able to fully meet those goals.

After the OKRs have been created, they should be published for other co-workers to see. Publishing the OKRs serves two purposes. The first purpose is to improve people’s performance with respect to their goals. Psychological research seems to suggest that people make better progress towards goals if they publicly commit to those goals. having other people know what you’ve set out to do improves your performance. The second purpose is to enable collaboration. It’s much easier to identify which co-workers you might want to collaborate with if you have visibility into which ones are working towards related or complimentary goals.

When a quarter or year comes to a close, and we can see how we’ve performed on our OKRs, what should we do with that information? Instead of using to directly drive formal performance reviews and compensation decisions, the book suggests using it to to do a more informal, low-stakes retrospective. The retrospective is intended to be an opportunity to pause and reflect on why one achieved the results that they did, and identify opportunities for self-improvement. That kind of frank assessment is unlikely to happen if OKRs directly tied to formal performance reviews, because the incentive then becomes to frame one’s performance in the most positive possible light, instead of taking a more balanced perspective on one’s successes and failures. If people see that OKRs directly determine their performance evaluations, then they become more likely to sandbag, which means to set excessively easy goals in order to reduce the likelihood of getting dinged for failing to meet a goal.

Even though the authors caution against using quantitative performance on OKRs as the sole criterion for performance evaluation, they are comfortable with taking it into consideration as one of several factors, alongside employee effort and goal difficulty. For example, one employee may make a poor effort but nonetheless fully satisfy 100% of their very easy goals, whereas another employee may put in tremendous effort and make significant progress towards a very difficult stretch goal, yet fail to fully meet all key results. Despite the second employee’s relatively lower quantitative performance on OKRs, they may deserve a better performance review than the first employee in light of the overall context.

Final Thoughts

Reading this book helped to deepen my appreciation for the psychological elements involved in successful goal-setting. It’s extremely important to fine-tune the details of how programs like this are implemented in order to ensure that incentives are aligned and people are motivated.

I would not recommend this book to readers with limited time because the ratio of meandering anecdotes to clearly structured suggestions is a little bit high for my tastes.