OKRs (Objectives and Key Results) seem to be the new flavor of the year for management and goal-setting. Recently, many articles, social media posts, and colleagues have been communicating about OKRs.  OKRs are "hot."

This framework started to emerge when Peter Drucker, the often-cited management consultant, introduced Management by Objectives (MBO) in 1954.  In the 1970s, Andy Grove, CEO of Intel at the time, improved MBO by introducing OKRs. In 1999, John Doerr, the renowned venture capitalist, introduced OKRs to Google. And since 2010, inspired by Google's example, many companies have been adopting OKRs. These days, OKRs seem to be the “new” thing in goal-setting and performance management.

But what is a “good” OKR, and how should we create them? Currently, many leaders, managers, and consultants are doing what they always do with “the next big thing" – recommending OKRs as a near panacea for management issues. There are training and certification programs for OKRs. If you look for guidelines for defining them, you often find the suggestion that they be similar to SMART goals (an acronym that, according to Smart As Hell guru Glenn Hughes has more than 100 different meanings). Perhaps the key elements are Specific and Measurable. One way to think about it is that we want goals to be specific enough so we can measure them.

Trained as a behavior scientist, I tend to be critical of many management and talent development tools because they are often too loosely defined, too abstract, or too open to interpretation. In behavior science – and I would say in performance engineering as well – we need clear definitions of the things we hope to improve so we can measure them.

The extreme case of widely adopted but unmeasurable descriptions of performance for performance management is competencies, which are abstract category names for clusters of behavior that can vary tremendously from one example to another.  Communication effectiveness as a competency, for example, looks quite different if it involves a tough negotiation with a potential business partner, selling a product to an interested prospect, or having a 1:1 conversation with a direct report. There are few if any similarities among the behavior needed in these situations except at the most abstract level. And yet, they all get clustered under a single label. Rating people on competencies is a form of refined opinion, not of objective measurement.

While OKRs are certainly better than competencies as a way to set goals, to measure performance, and to monitor progress toward goals, competency modeling sets a very low bar with respect to effectiveness. 

So, let’s look at OKRs a little more closely.

Describing Objectives in OKRs

An OKR is defined as an objective combined with a short list of key results used to measure progress toward the objective. Sounds good, but what does that really mean? If we search the Internet, we see a wide range of examples.  Most examples of objectives involve verbs, including:

• Increase efficiency of the QR process
• Delight customers
• Enhance security across the enterprise
• Deliver amazing customer support

Certainly, these statements of objectives are aspirational, and maybe even inspirational for some.  But until we know how to measure them, they don’t really mean anything. One might ask, “How long are you going to increase, delight, enhance, or deliver?”  In other words, when do you know you’re done, or successful?

That’s the problem with objectives stated as activity. They’re like having a meeting (an activity) that does not produce anything of value (e.g., decisions, plans, agreements, documents, etc.), but merely continues at the next meeting. It is not really a defined objective. At best, it’s a vaguely defined strategy or tactic intended to achieve an objective.  That is why we need the second part of the OKR – Key Results – to define the value of the activity described by the objective. We need something to anchor the objective in an outcome that is definite, measurable, and an indicator of success. 

Identifying Key Results

Here are some examples of "good" Key Results found on the Internet:

• Maintain a 100% service level for critical major consumables
• Redesign order process
• Increase Net Promoter Score (NPS) from 6 to 9
• Conduct a security assessment of code base using automated tools

Interestingly, these all seem to have verbs, as well – indicating an activity. When aggregate measures are used, such as Net Promoter Score (NPS), the result does not give me anything more specific to produce or achieve than an aspirational goal.

One might ask, what do we need to accomplish to increase the NPS?  What must we increase in order to increase the NPS to 9?  More customers who rate us 9 or 10 on the NPS scale? We can count them, much as we can count the number of people who rate a product at each level in reviews on the Amazon web site.  By identifying the countable accomplishment, we have made measurement straightforward.

Redesign order process, like some of the typical descriptions of objectives, begs the question, “How do I know when I’m done?” Some redesign projects result in worse outcomes rather than better ones. Might one at least suggest a cycle time or quality outcome to define the desired result?

The point here is not to attack these specific examples. There are examples all over the Internet, and they vary in all kinds of ways. That is because there is a lack of clear definition in the OKR guidelines about what constitutes an objective or a result.

Clarifying Objectives and Results Based on Performance Engineering

Let’s consider another approach for setting objectives and defining results with respect to performance, one grounded in the models and logic of Performance Thinking.

Performance Thinking® is a design engineering discipline. As with any form of engineering, it applies basic science. We know from behavior science that we can identify and count or tally specific forms of behavior, such as asking questions, writing messages, or assembling widgets. The common features of pinpointing behavior are to use a verb (action word) describing an action that can be repeated, and thus is countable. The underlying methodology for monitoring behavior in Skinner’s behavior science is to count instances of behavior and calculate rates of response, i.e., how many instances occur per unit time.

Extending to performance engineering, as defined by Thomas F. Gilbert, we know that the value delivered by behavior is not in the behavior itself, but in the products of behavior, i.e., widgets, decisions, documents, relationships, plans, agreements, prototypes, machines ready to go, etc.  In fact, behavior itself is generally costly in that it costs money, time and attention to enable people to behave on the job.

From a performance engineering perspective, we want to know what of value does the behavior produce, and how does the value of that product compare with the cost to produce it. This defines the second element of performance, the accomplishment or valuable product of behavior.  At The Performance Thinking Network, we call them work outputs. The definition of an accomplishment or work output is a product of behavior, expressed as a noun describing a countable thing not an activity. And to be valuable, it needs to contribute to organizational or societal results.

That, in the Performance Thinking framework, is the third element of performance: the results that define the success of a whole organization, or perhaps of a society. Organizational results might include revenue, profit, customer satisfaction, operational efficiency, safety, employee engagement, quality, employee retention, and so on. These are usually aggregate results contributed by many accomplishments, and they define the very success of the organization. While there might be intermediate operational results such as cycle time or sales efficiency, what we’re looking for here are the results that senior executives, business analysts, or owners equate with the success of the whole organization. These results are usually, in some form on their "dashboards."  Societal results are of a similar nature, but generally reflect a larger scope, such as a nation, mankind, or the planet as a whole. In any case, these are results that define success for the overall entity.

If there is no credible link from an accomplishment to overall results, direct or indirect, then that accomplishment may not be not worth spending the time, money, and resources to produce.

The Performance Chain model sums up these three elements, with definitions of each. We have found that when managers, leaders, and performance improvement professionals are precise about these definitions, their work can be far more effective because everyone knows what they are talking about. They can easily set goals and objectives, measure them, and know when they are achieved. We have also seen the implications of an accomplishment-based approach in other aspects of business and human performance management, as in our discussions with strategic planning master, Dr. Peter Dams, about making strategic plans more executable.

MBO, KPI, OKR, and SMART:  Murky without Clear Definitions of the Elements

Whether we are coaches, leaders, or performance professionals, we can frame conversations about goals by agreeing on what’s at stake for the whole organization. “Are we trying to improve profits, market share, customer satisfaction, or employee engagement…?”  Because knowing what we’re trying to achieve for the organization or society as a whole gives us both motivation and a north star to guide decision-making. We know that improving one or a few accomplishments might not move the needle on organizational results. But, we can usually determine fairly readily if a given accomplishment is likely to contribute to those results. We ask what’s at stake because that will shape how we proceed, influencing our decisions and priorities.

Notice that measures of organizational results are usually lagging indicators. That is, we don’t obtain data points very often needed to make data-based decisions. We might get organizational results measures once a month or once a quarter. We always want to be sure that a given increased value is not just an “up bounce” of variability that will go down next time, or the continuation of a trend that was already happening before we intervened.  In most cases, we need a minimum of 5 to 7 data points to be confident about trends and variability, and to thereby make truly informed decisions.

When setting goals, we can examine whatever area of performance we are discussing – whether for an individual, a team, a process, a business unit, etc. – and determine which important accomplishments comprise the value delivered in that domain, contributing to organizational results. These are countable things. We can define entire jobs or processes based on the accomplishments they deliver, and then select those of interest to incorporate into our goals. For an individual, it might be an accomplishment that needs improvement, one that is required for a next big project, or is the next step in a career ladder. For a process or a team, it might be the accomplishments that are holding us back, or the ones most critical to overall outcomes. For a business unit, it will almost always be accomplishments that will advance the organization’s results in a big way (e.g., a new team, a prototype product, new ideas about safety practices, good relationships with suppliers and customers, etc.).

Once we know the desired accomplishments, and “what good looks like” (agreed-upon, unambiguous criteria or standards that define when we have a good instance of the accomplishment), we can look in as much detail is needed at the activity or behavior needed to produce a given accomplishment.

It’s important to note, particularly given how often OKRs seem to include verbs to describe activity, that behavior for producing the accomplishment is neither the goal nor the result. Instead, behavior is how we achieve a valuable accomplishment. Part of working smarter or being strategic is to figure out the best, most effective, and efficient behavior for producing those accomplishments. If we focus on the accomplishments, then top performers might invent or discover better behavior for producing them, and we can learn from our exemplary performers.

In any case, this logic has the potential for dramatically clarifying KPIs, SMART goals, OKRs, or any other descriptions of goals and the means of accomplishing them.

Simply stated, a good goal would include what’s at stake for the organization or society and what we intend to produce or achieve that will contribute to that. We would likely state the goal starting with the accomplishment.

Accomplishment-Based Goals

Examples of accomplishment-based goals include:

• A user interface for our internal software system that user testing shows is quicker and more intuitive for employees to use, thus improving employee engagement and productivity. (With criteria based on user testing protocols for “quicker and more intuitive.”)
  
  
• At least 20 new introductions to CFOs who need our product, leading to sales and increased revenue (with a list of criteria that define what it means to “need” our product).
  
  
• Good working relationships with other software development team managers, to increase productivity and operational efficiency (with a list of criteria for what constitutes a “good working relationship”).
  
  
• A new Sales Enablement team that cuts across the usual silos for how we support our salespeople, for greater ROI and operational efficiency.

In short, a better way to define a goal is to describe a thing (an accomplishment) we want to produce or achieve, define what “good” looks like, and indicate how that thing will contribute to organizational or societal results.

Implications for Measuring Performance

Notice that when it comes to measurement, we can tally or count accomplishments – either over time (e.g., introductions per week), or when the one big accomplishment is complete. And we can break accomplishments into sub-accomplishments, either parts of the larger thing or things produced at each step toward completion of the bigger thing. So accomplishments that meet criteria for "good" are easy to count.

Usually, accomplishments are easier to measure more frequently, and thus give us a better leading indicator than most organizational results measures. We can usually count accomplishments more frequently than monthly or quarterly. Thus, they give us a better indicator to monitor the impact of continuous improvement based on measured accomplishments.  Whether you espouse SMART goals, KPIs or OKRs, they will be better if they are accomplishment-based.

-       - Dr. Carl Binder