OKRs Don’t Measure Human Performance
OKRs are a fantastic method of measuring business performance. But blindly applying them to measure human performance is a significant mistake.
First, some background:
What are OKRs?
OKRs are objectives paired with key results. First, we identify what we are going to do (our objective).Maybe our goal is to grow sales, or grow clients. Then, we specify how we are going to measure that growth (our key result). To do this, we must identify where we are now as a baseline, and what our target is.
A quick history of OKR success: OKRs started in the 50s with Peter Drucker and MBOs (managing by objectives). In the 70s Andy Grove used OKRs to propel Intel to remarkable growth and market leadership. In the late 90s John Doerr took OKRs to Google, and Google continues to use OKRs to drive growth.
Why do they work so well to drive growth?
Reason 1: Focus
When used correctly, companies have only a few top OKRs. These top OKRs are used for two things:
Determining what business goals to focus on to drive growth
Identifying work to stop that doesn’t drive the to the key result – even if the work seems valuable
For those of us in tech – we have seen this before. Imagine you have been working on a significant release, and there are code bugs to solve. You can exhaust thousands of hours solving bugs. OR, when you have a clear rule about what bug must be solved first, you can immediately identify repairs to focus on and others to ignore. (For example, when I worked with Walgreens, the top category of bugs was equivalent to, “There is a risk someone can die if we don’t solve this.”)
Stopping unnecessary work clears resources to focus on and accelerate key work that will make the most impact.
Nearly 10 years ago, I was with McDonald’s during the ‘year of the drive-thru’, and the rule of thumb was – if your project didn’t improve drive-thru business, it didn’t make the cut. Brilliant projects that you see in production now (kiosks for ordering and digital menu boards inside restaurants) were paused to put resources on drive-thru, which ultimately supported ~70% of McDonald’s US sales in normal times, and later, upwards of 90% during the pandemic’s peaks.
The bottom line – using the objective and key result to say ‘no’ is as important as using it to say ‘yes’.
Reason 2: Constant Measurement
OKRs create an expectation of constant experimentation and constant measurement. The best OKR structures I have seen are set quarterly, measured monthly, and transparently shared with the full organization. These frameworks show what is working, and what isn’t.
In my organization, we use transitional roadmaps. These roadmaps group product feature candidates that are prioritized and link to OKRs. These are features we THINK will drive the business metrics. We complete discovery (research that indicates the feature will succeed) and then complete testing (rapid Agile development on a small scale to see if the discovery holds true and we gain a measurable edge).
This means we fail quickly, and we succeed quickly – constantly determining within the Agile framework what development work to prioritize based on those results.
Both Focus and Measurement make OKRs exceptional for driving growth. They are not exceptional for measuring human performance.
Why you can’t use OKRs to measure people
There are at least 3 key reasons OKR fail to measure people and their performance well.
Problem 1: OKRs are company based, not team
Go back to the last section – did you catch ‘we fail quickly’? Not all the feature candidates are successful. Some teams build features that have a large impact on the Key Result, and others build features that have a lesser or no impact when their test fails.
If you use OKRs at an individual level to determine bonus, ratings, promotions – only the individuals on teams with top features in production will be promoted. Crazy ideas and fringe experiments have a higher rate of failure, yet as an organization we need those crazy ideas and those fringe experiments to uncover innovation. A hybrid ecosystem of features promoted to production and failed experiments is needed, otherwise we reward cautious, conservative ideas and miss the best chances to grow.
The holistic ecosystem itself (of tests that fail and others that succeed) and the overall outcome has to be rewarded.
Problem 2: There is necessary non-OKR work inside of companies
“Wait,” you say, “Isn’t the strategy that we only do work that ties to OKRs and we set other work aside?”
The secret is that there are critical, foundational things happening in companies that don’t tie to business growth strategy - bringing in talent, building leadership, closing the quarter, modernizing tech platforms, and so on. These bodies of work are about business strength, not growth. They are hard to measure (though some can be measured), and the resources we put toward them versus growth is a separate conversation. We could waste a lot of time trying to force these goals into OKRs, but I don’t believe that administrative overhead drives value.
Problem 3: Human talent development is an art form
All we see in research about human engagement from groups like Gallup is that the worst way to develop talent is to treat our talented people the same. Innate strengths, areas of individual curiosity, driving motivators – these are unique and when understood and optimized will drive exceptional outcomes from our people.
To translate this – if I am a great manager and developer of talent, I should be able to tell you a specific strength of each of my team members, what I am doing to help them build on it, and what I expect the value to the company to be.
You can be exceptional as a people manager or a contributor, ultimately driving material value for the company. It’s possible this management work doesn’t directly show up in an OKR (though it likely shows up in your development goals).
The solution:
Success comes from having 3 categories that make up our performance discussions, including compensation and promotion decisions:
OKRs – clear, few, transparent, measured often. When successful, they drive transparency, strong decision making and priority – and most importantly, results. These are usually broken down into nested Product OKRs that both support the top organizational OKRs, and serve to track all proposed engineering work / feature candidates in the Product & Engineering organization. Building Product OKRs and watching their results gives us stories and examples about employees that we can bring into talent roundtables. We can learn as much from how successfully someone deploys a key feature as we can from when someone stops a failed experiment, leads innovation brainstorming, and assesses the cost and value of features to prioritize them.
Reward structure for OKRs: Company bonus for all based on total performance.
Business goals – separate, these are usually set at a division level, and are confirmed by leaders as priority for business stability (modernizing platforms, improving retention, creating talent pipelines, maintaining security and so on). These are set by owners, and not necessarily transparent to all in the company.
Reward structure for Business Goals: Division / team or individual reward for owners of the work.
Human goals – separate, individualized. Aggressively personalized, these are the things we know drive engagement, excellence, innovation, and ultimately strong profits.
Reward structure for personal or human goals: Set individually, these clearly identify and build the case for promotion and opportunity.
This work we do is beautiful, challenging, and complex.
The structure for performance management needs to measure it all.