Which Key Result type to choose?

Category: OKRs

When you get started with OKRs, it seems like there is a lot to learn. Especially, how do you model key results? Is it even worth to go through the learning process?

Answer: Absolutely!!! It does take a bit of a learning curve. But once you cross that phase, you’ll wonder how we’ve been managing without these.

One of the most intricate ideas in the OKR world is how to represent your key results. We know that there are 6 types of key results. But which type do you choose in a given situation? The decision flow documented below is a simple guide to make that decision. It looks complicated, but it actually is not.

Let’s go through this process in a step by step fashion.

Is your key result something that can be measured or can it only be tracked?

Examples of measurables are revenue, customer complaints, employee satisfaction and so on. Examples of trackables are hire a sales manager, develop a sales plan and so on.

Let’s go down the path of trackables, since it’s a simple path.

Can you track the progress of your trackable key result in multiple steps or milestones?

A key result like “hire a sales manager” can actually have multiple steps:

  • Advertise the position
  • Screen and select top 10 candidates
  • Phone screen the 10 candidates
  • Select top 3 for in person interview
  • Select the top candidate
  • Offer
  • Onboard

In this case, we’ll take the “milestone tracked” key result path:

Let’s take another example, “Prepare product launch budget”. While you can drill down and come up with steps to track, you may be better off just indicating the status in terms of simple percent complete.

And in this case, we’ll take the “percentage tracked” key result path:

Now, let’s assume we go down the measurables path. We know how to measure what we are trying to improve.

Do you know the current (baseline) value of the KPI?

If you don’t know the current value of the KPI, you’d better want to know where you are today, so you can decide what to do next. In that case, we’ll take the following path:

What do you intend to do to the KPI?

Assuming that we know the baseline value, we’d want to do one of the following three actions to the KPI -- Control KPI, Increase KPI, or Decrease KPI. Let’s take the Control KPI path first.

What type of Control?

Maintain in range

Keep the value of a KPI in between a range of a minimum and a maximum value. For example, you want to maintain the among of training time that employee attends in a given period to be within 30-40 hours.

Maintain above a minimum value

Keep the value of a KPI such that it stays above a minimum value. For example, you want to release at least 4 articles per week to your blog.

Maintain below a maximum value

in your service to under 10 per month.

Let’s say, you’d like to increase the KPI, as an answer to Q4. Increase the KPI from the current levels to a new, higher level. For example Revenue, CSAT Score, Employee Satisfaction Score, and so on.

Are you tracking incremental increase, while the KPI actually starts from Zero in the new cycle?

Let’s take the “No” path. You actually don’t reset to “Zero”. For example, you want to increase the net promoter score from 60 to 70. When you go from one cycle to another, you start from where you left. In that case, it’s straightforward -- choose Increase KPI.

Now, let’s say “Yes” to Q6. You are trying to increase the revenue from $3M to $6M in the new cycle. If we model this as an “Increase KPI” KR, you’ll have problems checking in for a few periods because you need to build up the revenue to $3M to enter the measurement spectrum. In this case, it’s better to just say, model the key result to go from $0 to $6M and just use increase.

In some cases, this may not be possible. So, it’s better to divide the $6M by the number of check periods and use the “At Least” type of control KPI.

In our example, assuming a uniform distribution, we’ll say you have to achieve “At Least” $500K revenue every month.

Let’s say, you’d like to decrease the KPI, as an answer to Q4. Decrease the KPI from the current levels to a new, lower level. For example Customer Churn, Defect Score, Employee Attrition, and so on.

Are you tracking reduction from a higher value in the previous cycle, while the KPI actually starts from Zero in the new cycle?

Again, let’s start with the “no” path. You actually don’t reset to “Zero”. For example, you want to reduce “Customer Churn”, but that doesn’t reset to Zero. When you go from one cycle to another, you start from where you left. Again, pretty straightforward -- choose Decrease KPI.

Now let’s take the “Yes” path to Q7. For example, you want to reduce the defects in your production system from a total of 270 in the last period to 135 in the current period. A 50% reduction. But the defect count is essentially 0 at the beginning of the period. So, if you choose the “Decrease KPI” type of key result, your “good values” will never be in the spectrum since you want them to be always below 135.

In some cases, you may be able to work around this situation and still use the “Decrease KPI” key result type. But a more elegant approach may actually be to divide the 135 by say 12 months and try to create a “Control KPI” key result with a “At Most” sub type with a maximum value of say 11 per month.

Related Questions