"Who Does What by How Much?"

The secret to writing effective key results

Working with companies and teams implementing OKRs, one of the biggest questions I hear is, “How do we pick the right numbers or write the right key results?” Figuring out which things to focus on can trip people up—and it’s true that it is important to identify the right metrics so that your work actually makes an impact.

But writing effective key results for your OKRs doesn’t have to be hard. You just have to make sure you’re using key results that are outcomes, or measurable changes in human behavior that drive business results. That’s human, or customer, behavior—not system behavior. Teams often over-complicate or confuse the process by setting key results focused on system behavior instead.

Thankfully, there’s a great question you can use to make sure you’re focusing on human behaviors, and it’s one Josh Seiden and I like so much that we’re using it as the title of our next book:

[Who] [does what] [by how much]?

This is a slide Josh created that we use in our OKR workshops to show how the question ties directly to outcomes.

We like it because it puts the numbers into context by highlighting the humans (ahem, “who”) behind them.

So, how does our guiding question help clarify things exactly? How do you distinguish between system behavior and human behavior?

Human behavior v. system behavior

Key results have to have numbers. Looking at the data and analytics available, it’s all too easy for teams to think, “Great! We’ve got plenty of numbers here as options for our key results.”

For example, you might write this as a first key result: “Reduce the frequency of website pop-ups by 80%.”

It’s got a percentage (the number!), and it sounds like something that would help users.

But look closer: “Frequency of pop-ups” is a behavior of the system, not of the humans who use it. Setting that key result will help you determine to what extent you’ve implemented a feature, but it won’t tell you whether or not you’ve delivered more value to your customer.

This is where our question comes in. To clarify your key results, filter them through the question “[Who] [does what] [by how much]?”

The “who” has to be a human—in most cases, your customer.
“Does what” is the behavior you want to change in that human.
“By how much” is the measure by which that behavior needs to change (either increase or decrease) to achieve your objective.

Making sure the “who” is human

Let’s test it out with the example key result.

[Website pop-ups] [decrease in frequency] [by 80%].

The way the key result is written now, the spot designated for the “who” is filled with “website pop-ups”. Clearly, pop-ups are not humans, so how do we change the key result so that it focuses on the human behavior involved instead? Think about what the customers will ideally be doing if the pop-up frequency decreases.

[Our daily users] [leave our website within the first three minutes of their visit] [90% less often].

See the difference?

The first example is NOT a human action—it’s a feature of the website. The second example deals directly with “daily users.” Definitely humans. It may be that reducing the pop-up frequency will help you hit this key result; it may not. You likely have several options and routes to learn what exactly will help you change your customers’ behavior in the way you need.

The key to all key results

All key results you write should be able to fit within this structure. Remember, it’s the change in customer behavior that tells you you’ve done a good job, so it’s that change you need to target in your key results.


or to participate.