The Art of the OKR



Read the updated post, Art of the OKR Redux instead! A lot has changed since 2014…. 


Original “Art of the OKR” below, for archival purposes:

I wrote a book on using OKRs called Radical Focus: Achieving Your Most Important Goals with Objectives and Key Results Get it on Amazon or Gumroad

As well, I run OKR workshops to get you started if you want to adopt this powerful technique. Contact me.

The OKR approach to setting goals has been used at Google, Linkedin, Zynga, General Assembly and beyond and is spreading like wildfire across successful Silicon Valley companies.  The companies have adopted the approach are growing like weeds. OKRs provide focus, united the teams behind a single strategy, and makes all goals into stretch goals.  If  want to get your entire company to execute like the hounds of hell are behind them and the gates of Valhalla are open before them, try the OKR approach out.

OKR stands for Objectives and Key Results. The form of the OKR has been more or less standardized. The Objective is qualitative, and the KR’s (most often three) are quantitative. They are used to focus a group or individual around a bold goal. The objective sets a goal for a set period of time, usually a quarter. The key results tell you if the objective has been met by the end of the time.

Before you set OKRs, it is critical your company have a mission. Without a sense of purpose AND a scope to accomplish it, anything you do is equally ok.  I’ve written a bit on this in the North Star post.

But once you have a mission you a start attacking  it in a methodical way.  We do that by setting an objective.


Your objective is a single sentence that is

Qualitative and Inspirational
The objective is designed to get people jumping out of bed in the morning with excitement. And while CEO’s and VC’s may jump out of bed in the morning with joy over a 3% gain in conversion, most mere mortals get excited by a sense of meaning and progress. Use the language of your team. If they want to pwn it or kill it, use that wording.

Time Bound
e.g.  doable in a month, a quarter.  You want it to be a clear sprint toward a goal. If it takes a year, your objective may be strategy or maybe even a mission.

Actionable by the Team Independently
This is less a problem for start-ups, but bigger companies often struggle because of interdependence. Your Objective has to be truly yours, and you can’t have the excuse of “marketing didn’t market it.”

For example, some good objectives:

  • Pwn the direct to business coffee retail market in the south bay.
  • Launch an Awesome MVP.
  • Transform Palo Alto’s Coupon-Using habits.
  • Close a round that lets us kill it next quarter.

and some poor objectives:

  • Sales numbers up 30%.
  • Double users.
  • Raise a series B of 5M.

Why are those bad objectives bad? Probably because they are actually key results.

Key Results

Key results take all that inspirational language and quantifies it. You create them by asking a simple question “how would we know if we met our objective?” This causes you define what you mean by “awesome” “kill it” or “pwn.” Typically you have three key results.  Metrics can be based on

  • Growth
  • Engagement
  • Revenue
  • Performance
  • Quality

That last one can throw people. It seems hard to measure quality. But with tools like NPS*, it can be done.  With KRs you can balance forces like growth and performance, or revenue and quality by making sure you have the potentially opposing forces represented.

“Launch an Awesome MVP” might have KR’s of

  • 40% of users come back 2X in one week
  • Recommendation score of 8
  • 15% conversion

Notice how hard those are?

 KRs Should be Difficult, not Impossible
One great way to do this is to set a confidence level of five of ten on the OKR.  A confidence level of one means “never gonna happen my friend.” A confidence level of ten is also known as sandbagging. You are looking for a sweet spot where you are pushing yourself and your team to do bigger things, and where you have a 50/50 shot of failing.

Take a look at your KRs. if you are getting a funny little feeling in the pit of your stomach saying “It would take a miracle to hit all three of these…” then you are probably setting them correctly. If you look at them and think “we’re doomed” you’ve set them too hard. If you look them and think, “I can do that with some hard work” they are too easy.

What Makes OKRs Work?

OKRs Cascade

The company should set a OKR, and then each team should determine how their OKR leads to the company’s successful OKR. They can focus on a single key result or try to support the entire set. For example, engineering might decide satisfication is tightly connected with speed (and they’d be right.) So set an OKR like

Performance Upon Launch Equivalent to an Established Company

  • 99.8% uptime
  • <1 second response time
  • Instantaneous perceived load time (measure by survey, 90% users say page loaded “immediately”)

(I’m not an engineer so please do not mock my KR’s too hard)

As you can imagine, some teams like product management can easily align their OKRs with the company OKRS, while others may have to dig a little deeply to make sure they are supporting the company goal. Much of the value in OKRs comes from the conversations on what matters, how it will be measured and what it means for the teams who are used to working from their own standards, apart from the business goals.  Customer service, design and engineering often have to work a little harder to find meaningful OKRs that will move the business goal forward. But it’s worth doing. Can customer service upsell disgruntled customers to a better plan? Can design crate an onboarding flow that improves retention? Can engineer increase satisfaction with a better recommendations algorithm? No department can be an island.

As well, each individual should set individual OKRs that reflect both personal growth and support the company’s goals.  If the company’s OKRs is around aquistion, a product manager might decide she wants to “Get great at sales.” She then might chose KRs of completing sales training with a high score as well as improving the conversation rate of the product she runs.

Individual OKRs are about becoming better at your job, as well as helping your product get better. It’s also a gift to managers struggling with a difficult employee. In the Individual OKR setting process, she can work with that person to set goals that correct those problems before they blossom into full disciplinary actions.  By setting measurable KRs, she can avoid accusations of personal bias if things do not improve.

OKRs Are Part of Your Regular Rhythm

When people fail, it’s often because they set OKRs at the beginning of the quarter, and then forget about them until the end of the quarter. In between you are barraged by teammate requests, the CEO sending you articles, customer requests… there are always 101 interesting things to spend your time on that do not lead to success.  I highly recommend baking your OKRs into your weekly team meetings (if you have them) and your weekly status emails.  Adjust your confidence levels every single week. Have discussions about why they are going up or down.

Do Not Change OKRs halfway through your bounded time period.
Suck it up and either fail or blow past them, and use that learning to set them better next time.  No team gets OKRs perfect the first time. You are aiming to achieve 1-2 of the three KRs for you objective.  As well, changing them dilutes focus, and keeping teams focused is the entire point of the OKR.  Some companies have two or three OKRS, but I recommend starting (and perhaps staying with) only one company OKR, and one supporting OKRs set per team.

Get Ready to Fail… BIG!
Let’s be honest: we hate to fail. Everyone in the valley gives lip service to failure, but really we still don’t like it. But OKRs aren’t just about hitting targets but about learning what you are really capable of. They are there to push you to do better.  And honestly, if you missed the KR of 15% conversion but got 10%, are you heart broken? It’s a great metric, and you’ve learned a ton about what works and what doesn’t.

Compare that to the goal you thought you could make, maybe 2% conversion because you’d been reading the 80/20 articles. If you shoot for the moon, you may not make it but it’s a hell of a view.

Objectives and Key Results are a tested and powerful way to accelerate your company’s growth after you’ve found product market fit, or to focus the search as you explore.  I recommend them highly for anyone who seeks to accomplish extraordinary things.

Buy Radical Focus: Achieving Your Most Important Goals with Objectives and Key Results

Watch a talk on OKRs in action

My posts on OKRs and Execution

In a future post, I’ll answer some commonly asked questions about OKR’s. Feel free to ask your own in the comments.

…and others posts on OKRs
High Output Management by Andy Grove
How Google sets Goals: OKRs



* an approach originally laid out in The One Number You Need to Grow, NPS has been turned into a highly effective measurement of perceived quality. Startups, the short version is ask one question: How likely are you to recommend this to friends/family/colleagues on a scale from 1-10. if they say 9 or 10 you’re doing well. If they say 7 or 8 they aren’t talking. and if it’s less, they are probably badmouthing you.


Add Yours
  1. 1
    Mike Tobias

    Hi Christina,

    Thanks for the great post. It might be worth mentioning that OKRs is actually a very old idea based on Peter Drucker’s Management by Objectives and was implemented at Intel early on but called High Output Management. I know it’s not fashionable to say OKRs aren’t “new” and innovative, but it’s a boring old idea that happens to work very well.;)

    Thanks again.

  2. 2

    I don’t think I said they were new, just that they were tested and used at Google and Zynga and spreading throughout the valley. Sometimes a great idea takes awhile to take hold, especially in a new field.

    OKRs are based on MBO, and were introduced to Google, Zynga and others from the fabulous John Doerr who brought it from Intel. In order to keep the post focused, I decided to skip the full history (especially since it’s covered by the Google video I linked to).

    I do recommend Andy Grove’s High Output Management for anyone who wants to go deeper. However, there is not a lot on how to implement them effectively, prompting this post.

  3. 3
    Mike Tobias

    Quite right. I appreciate the focus on implementation. That’s the area where there isn’t much written; part science, part art. Fortunately, OKRs are spreading outside of tech as well.

    Thanks again for the great post!

  4. 4

    Great post, Christina. It addresses many points about implementation which are not addressed anywhere else.

    I feel certain that you say “Do Not Change OKRs halfway through your bounded time period” from your experience. I want to understand why. What would you recommend if an objective is no longer relevent for a team/individual a month after it was set (because of change in market/strategy/user response/new information not known previously).

  5. 5

    If you change, you are weakening the power of commitment. Do not make a commitment to a quarter unless you feel certain you can commit to that goal for three months. Pick OKRs that are appropriate to that time period. The OKR should be carrying you through surprises, changes and shifts and if they change every week or every month how is it acting as a compass? Keeping to it, even when you realize it’s not really going to work means marking up the effort a zero and starting again next quarter.
    If you give yourself permission to change, you won’t keep to the goal. As well, if you are entering a time of extreme change (like seeking product/market fit where picots come fast and furious), set an OKR that fits that time! How about an O of find product/market fit with KRs around acquisition, retention, conversion.
    It can take a couple weeks to get good OKRs that are robust. Take the time to make sure they are good, they guide, and they suit your team and your phase of life.

  6. 8

    Recent find:

    Creating quarterly OKRs (Objectives and Key Results) has been part of Google’s culture since board member John Doerr introduced the concept in 1999. More recently, however, we’ve elevated their importance and are using the quarterly OKR all-hands meeting (which is led by Larry Page and other company leaders) as a rallying point for all employees. Team by team, the leaders lay out their objectives and how they’ll measure success. Afterwards, they’re posted for anyone within the company to see.

    To make sure our products work seamlessly together across Google, we’re focusing more on broader OKRs – big company goals that can only be achieved if everyone works together. For example, a recent OKR objective for our search team was to improve the world’s information and make it universally accessible and useful, which restates and reiterates the company’s mission statement. The key results underneath that objective included metrics and projects for the quarter, many of which span a number of teams, ensuring a well-coordinated push toward a shared goal. Having these shared goals also has the benefit of helping prevent the formation of silos – always a concern as companies grow.”

Comments are closed.