Top/Company Level

We advise beginning OKRs coaching engagements with a mindset that your client will benefit by defining OKRs at the company level. OKRs at higher levels create context for OKRs at lower levels in the organization. To set OKRs at the company level, you must interface directly with the CEO. Assuming you have access to the CEO and can confirm that your client wants to define OKRs for the overall company, the company‐level OKRs become the top‐level OKRs. However, defining OKRs at the company level is not always the right place to start.

Your client may define a set of top‐level OKRs that exist below the company level. There are two reasons for this: (1) feasibility and (2) effectiveness. Feasibility is obvious: You may not have access to the CEO, especially in large companies with thousands of employees. The second reason is less obvious: Company‐level OKRs may not be effective. That is, the process of defining company‐level OKRs may not be of value. Many of our clients in the Global 1000 with hundreds of products and departments already have a strategic planning system in place that obviates the need for company‐level OKRs. In these massive organizations, introducing OKRs at a lower level often makes more sense.

Our clients have defined the top level as the office of the CIO, a business unit, and even a single product team. One of our larger clients with dozens of business units chose to define top‐level OKRs for each business unit. The company maintains an annual budget and metrics dashboard to capture the financials of each business unit and felt it would be a redundant exercise to define OKRs for the overall company.

Team Level

We recommend getting top‐level OKRs right before rolling out OKRs to lower‐level teams. This enables leaders to lead by example and may provide the context teams need to connect their OKRs to top‐level goals. If your client is eager to roll out OKRs across teams on day one, advise them to pilot OKRs with several teams prior to scaling their program. How your client chooses to implement OKRs at the team level is a critical predictor of their success with OKRs.

Your client's definition of team can make or break their OKRs program. When I worked at Betterworks, the leading OKRs software solution in 2013, we started every implementation by loading the org chart into our software. This automated the process of defining OKRs teams. Oh, wouldn't it be nice if we could simply define teams based on a one‐to‐one mapping to the org chart? While the org chart approach is simple and scalable, we find it is not always effective. In fact, we have encountered very few cases in which the org chart approach is effective across an entire organization over the long haul.

Dozens of organizations have approached us for coaching and guidance after concluding that their org chart approach failed. One of the most common complaints is that a functional team approach increased silo effects rather than alignment across teams. Defining team‐level OKRs based solely on the org chart often conflicts with the “work together” part of the definition of OKRs.

While no single approach works with every organization to define an optimal set of teams that should set OKRs, there are several approaches to consider based on what is proven to work in the field. Here are the three most common ways we've seen organizations think beyond their org charts to set up team‐level OKRs that encourage cross‐functional alignment. As you review each of the three approaches, keep in mind that they are not mutually exclusive. For example, your client might merge highly dependent teams, case 1, as well as define a cross‐functional team based on a top‐level OKR, case 3.

Case 2: Leverage Existing Cross‐Functional Squads    From 2014 to 2017, I was lucky enough to coach several leading online classified companies using OKRs in the United States, Canada, and the Netherlands. The online classified company in Canada completely abandoned the notion of functional team OKRs. They had an org chart with a clearly defined reporting structure. However, the org chart was totally decoupled from the teams that set OKRs.

Rather than a list of OKRs teams based on the org chart (e.g., product, engineering, human resources, finance, sales), the teams that set OKRs were mapped to the needs of various customer segments. This mapping reflected their cultural focus on customer centricity. Team‐level OKRs were associated with cross‐functional squads, each named after their customer segment. They included “Buyer Squad,” “Seller Squad,” “Auto Squad,” “Advertiser Squad,” as shown in Figure 3.1.

Figure 3.1 lists four squads defined by customer segments. However, this is just one example. One of our clients, a leading global e‐commerce company, defines squads based on different parts of their customer's journey. Its customer journey squads include: “Account Registration,” “Checkout,” “Payment and Fraud,” “Last Mile Delivery,” and “Recommendations.”

In his book Inspired, Marty Cagan describes a squad as a best‐practice approach for designing product teams:

A product team is a cross‐functional set of professionals, typically comprised of a product manager, a product designer, and a small number of engineers. In addition, there are sometimes additional people with specialized skills included on the team such as a data analyst, a user researcher, or a test automation engineer… . The key is that these people with their different skill sets usually come from different functional departments in the company, but they sit and work all day, everyday… . It's not unusual in larger organizations to have on the order of 20 to 50 of these cross‐functional product teams, each responsible for different areas, and each product team with its own objectives to work on.⁴

In 2017, we started seeing more and more organizations, especially in high tech, move beyond the org chart to define their OKRs based on cross‐functional squads. Leading organizations such as eBay, PayPal, Nike, and Walmart have implemented product squads. When such squads are in place, organizations find it intuitive to define OKRs at the team level for each cross‐functional squad.

Individual Level

If your client wants to set individual‐level OKRs, we advise making this practice optional. Other than Google, we have not seen an organization succeed when mandating individual‐level OKRs for all employees. Some organizations force individuals to define their OKRs by a certain date and load them into a certain location as a compliance ritual. Such practices introduce overhead and often create unintended consequences. Here's an anonymous story from a former customer success manager of an OKRs software vendor that illustrates some of these consequences.

One of our clients was an irate inbound call center manager. Her company had recently purchased OKRs software licenses for every employee and instructed all staff to publish their OKRs into the software tool by a certain date. She asked me how a given person working in the call center could possibly benefit by loading their personal goals into an OKRs software tool given that she already installed a call‐center tracking system that monitored every metric. She claimed, “No one at our company other than me cares about the individual performance of the workers in our call center.” Call center employees already had a real‐time dashboard that measured every metric that could possibly be worth tracking. The tracking tool cost quite a bit and was running well. We agreed with the call center manager that it would actually be a step backward if each of the nearly 200 call center workers were forced to copy and paste their metrics and load them into our software tool as a compliance exercise.

—Anonymous OKRs software customer success manager

We recommend that your client explore individual‐level OKRs only after succeeding with team‐level OKRs for at least two cycles. Even then, we advise organizations not to require that all employees define individual‐level OKRs, as this often turns an OKRs program into yet another compliance exercise. When individuals set OKRs for themselves, they often worry, rightfully so, that the achievement of OKRs will be used to evaluate their performance. As such, rather than setting stretch goals, they define key results with low targets so they can underpromise and overdeliver.

Yes, Google has thousands of employees setting OKRs at the individual level. While this approach has worked well for some workers at Google, we've spoken with Googlers who report that individual‐level OKRs are a waste of their time. In extreme cases, individual‐level OKRs become personal goals that are totally disconnected from business goals.⁵

We've met with a few teams at Twitter who abandoned the practice of setting OKRs at the individual level. These teams stopped setting individual‐level OKRs for two reasons. First, most OKRs at the individual level looked like a list of key tasks rather than key results. Second, some teams failed to make progress on team‐level OKRs, even though most individual‐level OKRs were achieved. In this case, the practice of defining individual‐level OKRs led to unintended consequences. Individual contributors tended to focus more on individual‐level OKRs rather than working as a team to achieve higher‐level OKRs.⁶ So, if individuals do not define their own OKRs, how do individuals connect their work to OKRs? How do individual contributors stay engaged?

Individual contributors can – and should – be actively engaged with OKRs at higher levels. Some individuals engage with OKRs by taking on roles such as team member, internal OKRs coach, and/or key result champion. Individuals often discuss OKRs during their regular 1:1s with their manager. Some use OKRs as context for focusing their effort on activities that drive OKRs forward. However, as OKRs do not capture all work, individuals also allocate their time to business‐as‐usual activities that maintain health metrics (e.g., update software to ensure website uptime remains >99.99%) or are simply required for compliance (e.g., provide reports to support the auditors).⁷

The guidance to begin with top‐level OKRs and to avoid requiring individual‐level OKRs is universal. However, your approach to choosing the teams that will set OKRs is perhaps the most important deployment parameter of all. This is an opportunity for your client to translate their corporate culture into the fabric of their OKRs program. Let's summarize three cases:

1. You want a hierarchical culture. Maybe you will choose the default approach to define OKRs teams based on the functional teams in the org chart. In this case, OKRs can help reinforce the org chart hierarchy.
2. You want a culture that fosters cross‐functional execution. Consider merging highly dependent functional teams to build your OKRs teams. This approach embraces dependencies and meets them head on. You may also define top‐level OKRs first and see if teams emerge as illustrated in case 3 in which our client created a cross‐functional squad to manage organic growth.
3. You want a customer‐centric culture. Consider defining team‐level OKRs based on the cross‐functional squad model as illustrated in case 2. Leverage this approach with clients that already have squads in place.

Binary Scoring (Oracle)

I first used OKRs in 2011 at a small startup company. We adopted a scoring system based on what worked for Oracle back in the late 1980s. Each key result was either achieved or not. Things were simple. It was binary. If a key result was “Sign 10 new customers by end of quarter” and the quarter ended with nine new customers, the key result was not met. In fact, it was assumed that you'd hit 10 customers midway through the quarter, cross out the 10, raise the bar to 15, and achieve 20 by end of quarter. This approach to key result scoring is known as “set the bar high and overachieve.”¹⁵

It was an unwritten rule that if your team achieved its objectives, your team would be celebrated, and individuals comprising the team would be more likely to be promoted. Your team was successful to the extent that OKRs were achieved.

Oracle's scoring system was not perfect. Consider the previous situation, where the key result was “Sign 10 new customers” but the quarter ended with nine new customers. The binary scoring system could create a sense of failure, as signing nine new customers was interpreted as falling short. Not by much, but still, the feeling was one of losing. Oracle's binary approach based on setting the bar high and overachieving was diametrically opposed to the culture of OKRs at Google, where the worst thing you can do is blow out all your OKRs.

Grading on a 0–1 Scale at End of Cycle (Google)

I learned about how Google grades¹⁶ OKRs right about the time the Google Ventures video came out in 2013. The idea was to standardize how all key results are scored across the organization. This normalized scoring model gives everyone a way of knowing how to measure success. A score of zero is bad, a score of 0.3 indicates some progress, a score of 0.7 reflects the target level of achievement, and a score of 1.0 reflects success beyond the target.

The culture at Google emphasizes stretch goals. In fact, scoring all 1s on key results implies gaming the system. I heard a story about a Googler who set lofty goals and then went on to achieve them all. Apparently, everyone assumed he sandbagged. Pranksters brought sandbags into his office as a practical joke.¹⁷ While it might not be perfect, Google's scoring approach standardizes conversations and streamlines communication about the level of achievement for each objective.

Stretch Target Commit with Predefined Scores (Lamorte)

Nearly all organizations score key results at the end of the cycle. Many score key results several intervals throughout the cycle. However, very few define scoring criteria when defining key results. Nearly all our clients find it valuable to establish scoring criteria up front. The conversation about what makes a “0.3” or a “0.7” is more meaningful when we translate these numbers into English. After discussing this with Vincent Drucker – and yes, Vincent is Peter Drucker's son – I developed the guidelines shown in Figure 3.2 that our clients find useful.

Here is a story that illustrates the value of aligning on key result scoring up front. The story begins with a VP of engineering creating the following OKR:

The key result as written represents the 1.0/stretch scoring level. Aligning on the 0.3/commit and 0.7/target scoring levels before finalizing key results forces a conversation about what is a commitment versus what is a stretch (or aspirational) outcome early in the OKRs cycle. When presented with this scoring model, some of our clients are initially concerned that defining three levels will require too much time. However, in practice, defining these scoring levels takes just minutes.

In this case, the engineering manager let the VP know that the drafted 0.7/target score was not even possible and that the 0.3/commit score would be quite a stretch. The engineering manager had a contact in the legal department who explained that the organization did not have permission to even sell the product until the following quarter. Surprisingly, the VP was not aware of this legal barrier.

The prescoring exercise aligned expectations. It enabled the VP to learn that it would be impossible to get paying users by the end of the quarter. Having these conversations before finalizing the key result ensured everyone was on the same page from the start. In this case, to better manage expectations and resources, the engineering team refined the key result. Ironically, the proposed commit level of the key result as drafted by the VP ended up as the stretch level of the finalized key result:

Even though Google uses a 0–1 numerical scoring system, most of our clients do not adopt a numerical scoring scale. Rather than using 0.3, 0.7, and 1.0, they simply use words such as commit, target, and stretch to align on three levels of progress for each key result.

Mid‐Cycle Scoring: Historical versus Predictive

Many organizations that approach us for help with OKRs have already started working with OKRs and have implemented a scoring system that focuses on progress to date. As such, they measure historical progress on each key result in the form of “X% complete.” While we do not discourage you from looking back and monitoring historical progress on key results, we highly recommend including a predictive element. Let's go back to the “10 new customers” key result to analyze why predictive scoring is so valuable.

Suppose we sign six customers in the first month of the quarter. Great, the key result is 60% complete! However, suppose this same team is not confident that they can sign 10 total customers by quarter end; perhaps the pipeline dried up or a key sales rep just resigned. It would be nice to have a scoring system that communicates this concern.

Predictive scoring serves as an early‐warning system to better manage expectations and reduce the surprises many leadership teams would rather avoid. The simplest form of predictive scoring features a single confidence level that a given key result will be achieved or not by the end of the cycle. Radical Focus adopts such a system. All key results begin with a confidence score of 5/10 by definition and scores for each key result are updated weekly throughout the cycle. In this case, we could communicate that the confidence score dropped from 50% to just 20% to reflect the unlikelihood of signing 10 customers by quarter end. It is the change that then triggers conversations about progress.¹⁸

Predictive Scoring That Goes beyond the Numbers

Our analysis of scoring thus far has focused on the actual achievement level of the key result. However, the numerical score of a key result often does not tell the full story. For example, your client may be on track to achieve a key result to onboard five new customers, but achieving the key result requires excessive overtime. The key result champion may provide a high confidence score but feel like saying, “We're on track to achieve the key result, but . . .” If there is something that feels bad about the key result, they could flag this by coloring the key result red or selecting a sad‐face emoji to escalate and resolve as a team. While some systems use color coding based on numerical scores, other organizations introduce color coding to look beyond the numerical progress of a key result.¹⁹

One of our clients had a key result to get 100,000 mobile app downloads. They were on track to get 200,000, but the quality of these downloads was horrible. In fact, they were gaming the system by purchasing low‐quality downloads to achieve the key result. The key result was not driving the right behavior. Given that the quality was so low, the key result scored a “red 1.0.” The opposite case of a “green zero” can also occur.

In the case of a “green zero,” there is no progress on the numerical score of the key result, but there is a positive development. Consider the key result, add five new customers this quarter. If your client is on track to add zero customers this quarter, the key result scores a zero. However, they might have 10 new customers lined up to sign next quarter and feel compelled to color the key result green.

Systems for scoring OKRs are still evolving. We predict that the topic of scoring key results will become more widely discussed as organizations hit a roadblock due to their lack of a clear approach to key result scoring.

Analysis of the Three Most Common Scoring Systems

In the Radical Focus scoring system, all key results are set at a 5 in 10 confidence level. It is expected that each team setting OKRs will achieve roughly half of its key results. Key results are updated weekly with a predictive confidence score using a 1–10 rating. Any change in confidence level should trigger conversations. At the end of the period, each key result is either “achieved” or “not achieved.” If you achieve 80% of a given key result, it receives a final score of zero. We find teams often achieve “all or nothing.” That is, some teams tend to achieve all their key results and others tend to achieve none of them. Teams learn to adjust their level of stretch to better manage expectations as they define future OKRs. Radical Focus scoring often works well for organizations just getting started with OKRs that demand simplicity.

The Measure What Matters scoring model requires each key result to be classified as a “commitment” or an “aspiration.” At the end of the period, each key result is given a final score, often called a grade, between 0–1. Although this system encourages teams to set both commit and aspirational key results, most teams emphasize commit key results. In our experience, teams adopting this system tend to create commit key results that look like a lengthy to‐do list. Aspirational key results, if any are included, are often added as an afterthought. With most key results written as commitments in the form of binary milestones, the scoring process can create confusion. We once saw a final score of 0.4 for the aspirational key result “End Q3 with a permit to build in Portland.” What does that even mean?

In the Stretch Target Commit approach, all key results are written as stretch outcomes that are typically defined at a confidence level of 1 in 10. The same key result also has a predefined commit and target level. The commit reflects a 9 in 10 confidence level and should be highly controllable with minimal dependence on external factors. The target level reflects a 5 in 10 confidence level, falling somewhere between the commit and stretch.²⁰ While we find this approach works best in most cases, some organizations benefit by adopting one of the other scoring systems. Figure 3.3 compares the three most common scoring systems.

Principle 1: OKRs Should Be Included in Performance Review Discussions via Structured Questions to Position Managers as Coaches

Given that team‐level OKRs reflect a focus for improvement, it is only natural that individual performance reviews incorporate a structured discussion of their team's OKRs. We advise our clients to consider incorporating OKRs‐related questions into their performance review process. These questions are informed by why they have chosen to implement OKRs. In addition, managers may bring these questions into 1:1s with their direct reports.²⁸ Here are sample questions our clients have adopted to address impact, focus, communication, and learning:

• Impact. Which key result/s do you feel you impacted?
• Focus. How did you use OKRs to focus on high‐priority work?
• Communication. How did you leverage OKRs to better communicate?
• Learning. Which key result did you learn from? How will you apply this learning going forward?

While some organizations include OKRs questions in their formal performance reviews, others distinguish between a performance review 1:1 and an OKRs review 1:1. One of our clients already had a quarterly performance review process in place. They happened to define a four‐month OKRs cycle time. This client stumbled into two unexpected benefits resulting from these distinct time frames: (1) it was easier to decouple OKRs from the performance review process; and (2) an increase in the number of structured 1:1 conversations from four to seven each year.

Managers can certainly have more than seven such 1:1s over the course of the year, but these seven touch points are required and documented with HR. While there may be some overlap between OKRs and performance management, there is an agreement to address OKRs and performance management in separate meetings. This approach may resonate with your client if they are looking to create more engagement between managers and their direct reports.

Principle 2: Scores on Key Results Should Not Be Used to Calculate Incentive Compensation

Incentive compensation should be calculated independently of OKRs. Some key results may not connect to incentive compensation at all. Other key results may correlate with bonuses. For example, the value of a metric key result based on revenue is often used as part of a bonus calculation. That's fine. It is the value of the metric key result that determines compensation, not the score. The distinction is subtle, but important. OKRs should enable us to think big as we align on stretch goals. Linking key result scores to compensation creates incentives to set low targets and may reinforce silo thinking. Two examples are:

Example 1: Key result value (not score) tied to bonus

• Sales key result: Increase new sales from $1,000,000 to $2,000,000 this quarter (prescoring: commit = $800,000; target = $1,200,000)
• In this case, the sales team is paid a bonus based on revenue from new sales, so the actual value of this metric is directly correlated to the bonus. However, bonuses are not determined based on whether a score of commit, target, or stretch is achieved. The bonus is calculated based on the value of the actual metric itself.

Example 2: Key result not tied to bonus

• Human resources key result: Increase the number of phone screens for VP candidates from 40 last month to 80 this month (prescoring: commit = 40; target = 60)
• In this case, there were no bonuses calculated based on the number of candidates screened over the phone. This is still an important metric to the organization, but neither the scores on the key result nor the value of this metric determine bonus payout.

High‐Level Overview: OKRs versus KPIs

Based on our experience with hundreds of organizations that use KPIs, we can tell you one thing about KPIs for certain: the term “KPI” does not have a standard definition. Some companies use KPI interchangeably with metric. That is, every metric that is reported is also referred to as a KPI. While some companies have just one KPI, others have thousands.²⁹ Some organizations use KPIs as the basis for calculating incentive compensation for all staff while other organizations define KPIs to evaluate performance of the overall company and do not even offer incentive compensation.

Unlike KPIs, key results do have a standard definition. Key results answer the question, “How will we know we've made measurable progress on a specific objective by a certain date?”

Organizations with an established KPI system in place often ask if they should replace KPIs with OKRs. The fact that they are asking this question reflects a deep misunderstanding. Choosing between OKRs and KPIs is a false choice. OKRs and KPIs work together in tandem. A given KPI is a key result if it is the focus for near‐term improvement. A KPI is classified as a health metric if it is important to monitor but is not the focus for near‐term improvement. Metric key results are typically based on underlying KPIs.

To illustrate how OKRs and KPIs work together, consider a company with the objective: “Achieve financial targets” and the following three key results:

1. Double company revenue from $5M in Q1 to $10M in Q2.
2. Increase gross profit margin from 20% in Q1 to 25% in Q2.
3. Increase recurring revenue from existing install base from $400k in Q1 to $600k in Q2.

Each of these metric key results has the KPI in bold built in. In fact, metric key results equate to moving a given KPI from X to Y within a set time frame.

At the team level, milestone key results are more prevalent. Milestone key results do not translate directly to KPIs. Differentiating between metric and milestone key results helps clarify how OKRs and KPIs work together. Suppose a marketing department has the following OKR:

Notice that marketing cost per lead, in bold, is a KPI at the marketing team level. However, “sending our first set of nurture emails” is unlikely to be classified as a KPI; this statement is classified as a milestone key result. Although a milestone key result is not defined as movement of a numerical KPI from X to Y, milestone key results should be designed to impact a KPI in the future. For example, achieving the milestone of launching a marketing automation system may not impact marketing cost per lead in the current period. However, it is expected to have an impact on a KPI in the future, as illustrated in Figure 3.4.

Detailed Analysis of KPIs versus OKRs

The original question of how OKRs differs from KPIs is better framed as: “How do key results differ from key performance indicators?” Objectives are qualitative statements, and as such, should not be confused with quantitative KPIs. To distinguish between KPIs and key results, some of our clients like to include a chart like the one shown in Figure 3.5 in their OKRs training materials.

Defined in Context of an Objective?

The most obvious difference between key results and KPIs is the fact that key results are defined in the context of an objective. The objectives that contain key results should include a brief analysis of why the objective is so important now. While KPIs are sometimes defined in context of higher‐level goals, they often lack context and appear as a list of metrics to quantify the performance of some part of the organization.

Achievement Tied to Compensation?

While KPIs are often designed specifically to determine incentive compensation structures, OKRs are not. OKRs should be decoupled from compensation. While scores on OKRs should not be used as the basis for determining incentive compensation, OKRs are often incorporated, qualitatively, as part of the performance review process. The most critical reason for separating OKRs from incentive compensation goes back to Andy Grove's original intent that OKRs should be stretch goals.

Visible to All Employees?

Management by Objectives (MBOs), Andy Grove's inspiration for creating the OKRs model, is often used as a system for determining incentive compensation. Thus, MBOs are often kept secret between manager and employee. Similarly, KPIs are often tied to compensation and may therefore be kept private between manager and employee.

Andy Grove wanted to take the MBO model to the next level. Grove insisted that OKRs be made visible across the entire organization. The fact that everyone can see a set of well‐defined goals in a standard format sets the stage for alignment. With OKRs, the question is not whether to make them visible. The question is to what extent.³⁰

Most coaches advise their clients to make OKRs visible across the entire organization. However, some of our clients begin their OKRs programs by giving access to OKRs only to certain groups of employees, for example, director and above.³¹ Given that nearly all organizations decide to make OKRs visible to all staff by default, we do not consider this a universal deployment parameter. Nonetheless, as some organizations choose to specify restrictions, we've added visibility to the list of “other deployment parameters” at the end of this chapter.

Focused on Maintenance Work and Health Metrics?

Many organizations classify every metric they measure as a KPI. However, the fact that you are measuring something does not make it a key result. If we think of KPIs as the set of metrics a company measures, then a KPI is either a key result or a health metric.

A given KPI is classified as a key result if the company decides to focus on improving the value of the KPI in the near term. If a KPI is not the focus for near‐term improvement, it is classified as a health metric. Health metrics are already within an acceptable range. For example, while website uptime may be a KPI, if the website is currently at 99.99% uptime, it is unlikely to be a key result that is a focus for near‐term improvement. Does it really make sense to define a key result to drive a team to “improve website uptime from 99.99% to 99.999%”?

Intended to Increase Alignment across Teams?

KPIs are often created to measure a given team's performance. Team members care a lot about their own team's KPIs. However, they are often disinterested in the extent to which other teams achieve their KPIs. In fact, a team member might even get a feeling of superiority when their team achieves its KPIs and other teams do not. Conversely, the OKRs framework seeks to get employees working together to make measurable progress. OKRs are often defined and shared across functional teams. As noted in the first deployment parameter, defining OKRs may even lead to the formation of cross‐functional squads.

Informs Near‐Term Focus and Prioritization?

KPIs do not require a specific time frame. They are simply metrics such as “revenue,” “room nights booked,” and “net promoter score.” KPIs may or may not have target values within a set time frame. As KPIs include health metrics, lists of KPIs can become massive. KPIs are often intended to be reported in dashboards rather than to communicate the near‐term priorities.

Conversely, key results articulate how an organization will measure progress of an objective within a set time frame. While KPIs tend to stick around year after year, most key results are modified or removed as part of the reflect and reset step in the OKRs cycle.

Controllable by a Single Team?

Given KPIs are often used to evaluate a team's performance, it is only natural that KPIs tend to be totally controllable by a single team. However, key results are often written as amazing outcomes. Amazing outcomes tend to depend on external factors that may not be totally controllable by a single team. And that's fine. Both Measure What Matters and the Stretch Target Commit scoring systems define “commit” key results that should be mostly controllable by the team. Nonetheless, all scoring systems, including Radical Focus, advocate setting key results that stretch teams out of their comfort zone rather than what they feel is completely within their control.

Originates from Team Members?

As KPIs are often used to measure performance and even determine bonuses, they tend to come from the top. While most objectives come from the top, most key results originate from team members rather than the boss. To clarify, some key results will (and should) come from top‐level leadership. However, the process of creating key results involves a dialog between team members and leadership. This leads us nicely into our last two universal parameters where we explore how to align OKRs and ensure that most key results originate bottom‐up.