Culture / Technology

A Guide to OKRs and Overcoming the Pain of Them

27 Jan 2022 3:00am, by
three measuring rulers

A lot of managers swear by OKRs — objectives and key results — as a measurement of employee performance. A lot of employees just swear at them.

OKRs are a divisive measurement in the tech industry, a true love-hate experience. Either you find it really motivational to have shared goals that connect back to your organization’s, or you are frustrated you have to set goals that you only have a 70% shot of hitting.

While the origins of OKRs date back to management influencers Andrew Grove and John Doerr in the 1970s, the concept didn’t really make a mark until Google started experimenting with them in 1999. Later, the likes of LinkedIn, Twitter, Uber and Microsoft got on the bandwagon.

Still, OKRs didn’t blow up until Larry Page, co-founder of Google, wrote a forward in Doerr’s 2017 book “Measure What Matters,” in which Page said OKRs drove tenfold growth at Google.

But what works for Google won’t work for every company.

“Most of OKR initiatives are, unfortunately, failures. There’s nothing simple about OKRs. The only thing simple is the theory. It’s a bit like swimming,” Almudena Rodríguez Pardo, a business agility consultant, told The New Stack. The theory of swimming is pretty easy to grasp, but that theory alone won’t keep you from drowning.

With help from Rodríguez Pardo and other experts, we try to get to the bottom of how OKRs work best and how to overcome their all-too-common pitfalls.

What Are OKRs?

OKRs offer an alternative to key performance indicators (KPIs), which typically have performance reviews, promotions, salaries and bonuses hinging on meeting at least 100% of the goal. OKRs are made of two parts:

  • Objective: an unmeasurable goal.
  • Key results: two to four time-bound metrics that support each objective.

While KPIs are meant to be attainable, both your objective and your key results are meant to be aspirational. That’s why you set key result targets at about 30% higher than you’ve already achieved or perceive you ought to achieve.

Setting such stretch goals means if you make, on average, 70% of your key results, you’re likely to have moved toward your objective and you’ve probably set the right goals. If you hit, say, only 50%, you need to rethink whether you’re setting your sights too high or need extra support. And if you meet 100% of your key results, you’re likely not being ambitious enough. Sometimes the simple act of trying harder to reach objectives will result in exceeding expectations.

“I worked on a team with an already really high NPS [net promoter score] of over 70. A new team leader came in and asked us to go for 80. We hit it. Then she said we should shoot for a 100 in our OKRs,” which Leanne Creasey, a leadership development team lead, and her whole team, thought was ridiculous.

But then they worked harder to prove their new boss wrong. “We created an objective to engage with customers about NPS, asking questions like: Why did you score us that way? How could we do better? What would a perfect score look like for you? End of quarter, we didn’t get to 100, of course, but we went up to 89.”

Like KPIs, OKRs work in layers:

  • Organizational OKRs
  • Team-based OKRs
  • Individual OKRs

This aims for a balance of top-down vision-making, and team and employee autonomy. While KPIs are often kept to the executives, OKRs should be very public within a company in order to increase the transparency that drives alignment.

Too many companies, Rodríguez Pardo argued, rush to set individual and team-based goals from the bottom up before they’ve succinctly communicated the C-suite’s vision for the company (“Just a tweet, no Powerpoint, please,” she advised) and the strategy to get there. OKRs must cascade from the top down, she added, because they are not about success, so much as making sure everyone is heading in the right direction.

Vision: Conquer the galaxy! Objective A: Kill All Jedis Key Result 1. Identify all jedis Key Result 2. Murder all jedis Measurements 1.1 More than 98% Jedis spotted 1.2 More than 98% Jedis eliminated Objective B Wipe all rebels out Key Result 1. Trace all rebel camps Key result 2. Destroy all rebel camps. Measurements 1.1. more than 80% camps identified before June 2022 1.2 more than 80% camps destroyed by December 2022 Objective C Powerful Destructive Weapons Key Result 1: Develop the Death Star Key Result 2: Lightsaber 2.0 Measurements 1.1 Two hostile planets wiped out 1.2 Lightsaber twice as more powerful as version 1.0 1.3 Lightsaber 2.0 same weight as 1.0

To explain top-level OKRs, Rodríguez Pardo turned to a “Star Wars” analogy. What’s the  Galactic Empire’s clear vision?: “To conquer the galaxy.” One of its three top-down objectives is to wipe all rebels out. From this comes the key results of tracing and then subsequently destroying all rebel camps.

The measurable results to work toward are then identifying 80% of rebel camps by June, followed by the destruction of 80% of rebel camps by the year’s end. Ideally, the OKRs would be set quarterly and then reviewed at the team level at the end of each sprint.

This is not to say that the Empire made the right choice in how it looked to conquer the galaxy — it could’ve chosen negotiation or civilly divvying up planets — but this violent strategy had a clear focus and systemic view that was certainly successful for a while. And pretty much all employees, excepting Finn, were aligned — or resigned — to it.

Are OKRs Measuring the Right Things?

OKRs are typically found at organizations that heavily value continuous improvement, which can only be achieved through continuous measurement.

Still, OKRs have been notoriously a struggle at many organizations. Perhaps because they are KPIs or highly linear Waterfall or micromanagement in sheep’s clothing. Or perhaps old habits die hard and what’s being measured isn’t what really matters.

Mik Kersten, founder and CEO of Tasktop and author of “Project to Product,” spoke at 2021’s DevOps Enterprise Summit about tracking DevOps metrics using both OKRs and flow metrics. From the start, he offered a big promise that OKRs, when done right, can take teams from a project to a product mindset, and from ticking off tasks to tracking outcomes.

“OKRs speed our decision-making. They help us pivot and make adjustments faster. They help us experiment and test hypotheses faster,” Kersten said.

But they require a change in mindset, he added. “The challenge is when we snap into old behaviors while renaming them with OKRs.”

Kersten’s company produced Tasktop Viz, a value-stream management platform, which allows for tracking of KPIs and OKRs. He shared some of the metrics from data collected over the past year:

  • Only 8% of what’s planned by Agile teams gets delivered, showing a huge capacity mismatch.
  • Twenty percent of features were canceled after code was written.
  • Thirty-five percent of products have zero capacity for new work over the next year, so why are they setting new outcomes?
  • Eighty-five percent of products under-invest in security and technical debt.
  • Ninety-five percent of organizations don’t know the true efficiency rate of their value streams.

Often, both efficiency and capacity are not being communicated to the business stakeholders, Kersten found, so OKRs are futile.

Most OKRs, he said, still focus on business and financial key results, which are important. But only focusing on those factors will stunt growth. Innovation is equally important for tracking outcomes for customers.

And OKRs don’t have to replace KPIs. In fact, these indicators can measure the health of a business. As with a car, it’s good to keep these indicators in a certain range, Rodríguez Pardo said. If you don’t have petrol — or, in the case of a business, enough revenue to pay staff — you can’t get going, no matter which way you’re facing.

The Problems with OKRs

One hazard of faux OKRs is that Waterfall planning returns, and multilevel activities just become top-down, long-form task dictation. This is why you should limit each level to a maximum of four objectives, and four key results for each.

OKRs should be about outcomes, not activities. It’s less of a project mindset — do this, then that — and more of a product mindset — how are you delivering outcomes to your team, stakeholders and customers? Like all successful modern organizations, they must include both business and technical goals.

OKRs have important consequences, including deciding what not to do. Rodríguez Pardo gave the example: her own consultancy’s 2022 OKRs are to expand in Western Europe and Mexico. She was invited to give a talk in Lithuania, a country she’s longed to visit, but because that deters from company objectives, she declined.

She pointed out another OKR pitfall is setting them as departmental goals. For OKRs to be a success, they must have the DevOps focus of breaking silos. So often, developers’ work is distant from the company goals, which Rodríguez Pardo called “The Jira Problem”:

  • What are you working on? User story 748.
  • What’s that about? To make feature X.
  • But why are you making that feature? How does it support the overall strategy? Errr…

A modern organization understands that no project is done in isolation. HR may need IT to make a tool to help it successfully recruit. For the Galactic Empire’s objective, team-level goals to support the conquest of the galaxy included spies, drones, and AI technology and prototyping to feed into more advanced weaponry.

Following the Vision from the previous image, it takes down to a team-based level, adding brainstorming, including images of teams around whiteboards: Objective A: Kill All Jedis Key Result 1. Identify all jedis Key Result 2. Murder all jedis Measurements 1.1 More than 98% Jedis spotted 1.2 More than 98% Jedis eliminated Spies? Lightsaber 2.0? Objective B Wipe all rebels out Key Result 1. Trace all rebel camps Key result 2. Destroy all rebel camps. Measurements 1.1. more than 80% camps identified before June 2022 1.2 more than 80% camps destroyed by December 2022 Drones? Spies? Existing fighter planes? Objective C Powerful Destructive Weapons Key Result 1: Develop the Death Star Key Result 2: Lightsaber 2.0 Measurements 1.1 Two hostile planets wiped out 1.2 Lightsaber twice as more powerful as version 1.0 1.3 Lightsaber 2.0 same weight as 1.0 AI technology? Prototype?

Some experts, like Jeff Gothelf, an IT consultant and author, argue against creating individual OKRs completely.

“Setting individual OKRs generally leads to goals that are either not true indications of meaningful progress or that are easily gameable,” he wrote.

Individual OKRs tend to be tasks (like taking three coding courses in Y language), instead of key results (like measurably improving coding skills). Or, if the goal is to reduce bugs, the developer will take less risks to try to meet that goal.

In this world of group goals, how do you hand out bonuses? How does this work on teams like sales, which are accustomed to individual commissions? The most important downfall of OKRs is when compensation is tied to OKRs, according to Rodríguez Pardo.

“The moment we put money on OKRs, we pervert the systems — give me a number, I will make it,” she said.

Sales is a good example of how individual objectives can lead to competition and even aggression. Often, the consultant said, when working with telecommunications organizations, she saw sales so siloed that the sales personnel would sell things that IT couldn’t deliver in time, which would create a backlash from customer support.

Anyway, she asked, “​​Who did you hire if you have to put pressure with money for people to perform?”

Whether you decide to do multiple levels of team OKRs or go all the way down to individual ones, that top-level vision must be highly visible and constantly discussed. An OKR SaaS tool is an easy way to facilitate getting each and every employee on board and aligned, but don’t get distracted by focusing on dashboards over conversations.

Finally, Rodríguez Pardo warned that many organizations have way too many OKRs. Don’t even kick off with individual ones until you’ve nailed them at the top and team levels. She recalled an organization she was called to consult with that had 600 people working on 800 objectives: “If people are too busy with OKRs, they won’t actually do work.”

Managing for Flow

“If I can state one challenge around the way OKRs are being deployed today, it’s that they ignore the improvement of daily work,” Kersten said.

Backlogs are already overflowing. Tech burnout is almost a given. And no capacity is being allotted to improve daily work.

Above all else, OKRs are meant to establish faster feedback loops toward continuous improvement. But Kersten’s team has uncovered certain anti-patterns that dramatically slow that down:

  • Using OKRs as a tool to micromanage value streams.
  • Waterfall planning via multilevel tasks.
  • Focusing only on business and financial metrics.
  • Applying team-level telemetry — story points, uptime, velocity — to organizational key results.
  • Feedback cycles are too slow, more than 90 days.

In general, he says this all comes back to an organization not adopting the three ways of DevOps:

  • Flow.
  • Feedback.
  • Continuous learning and experimentation.

Improvement needs to be measured. Kersten has found that the key results are quite simple to understand because they are the things the majority of businesses are already measuring — net promoter scores (NPS), retention, and financial gains and losses.

But, on the whole, he found that most teams couldn’t find ways to track progress, like to see if DevOps initiatives were actually enabling teams to achieve more of their key results.

Kersten discovered a need to create flow key results, in order to identify bottlenecks that make delivering business key results harder. He developed the Flow Framework to identify four flow categories for improvement:

  • Features: new business value, based on data from customers.
  • Risks: security, availability, compliance, based on data from risk officers.
  • Defects: quality improvements, based on data from customers.
  • Debts: technical debt improvements, based on data from IT teams.

These are “not mutually exclusive and comprehensively exhaustive,” he said — understanding, for example, that feature velocity in isolation can contribute more bugs, security flaws and technical debt.

After all, OKRs are inherently cross-functional and, as Kersten put it, “everyone working on that particular value stream needs to understand what outcomes they’re driving, who the customer of those outcomes is.”

Getting Started with OKRs

In 2019, when Alister Harris and Mark Chamberlain acquired the 20-year-old enterprise software company now known as Lokulus, they were looking to shake things up. One step was to build a new HQ with no doored offices, even for leadership. Another was to move from “old-fashioned” Waterfall annual goal-setting to something more fast-paced with more focus.

“When you’re setting annual goals, you’re too far away from them for nine months a year,” Harris said. The company was looking for a framework that better aligned the 50-person team toward a singular objective.

He dubbed the company’s original four-line vision statement as “hyperbolic” — something immeasurable about being an innovative company. The organization moved to OKRs because, he said, “they are simple and rememberable. One focus, one drive.”

Lokulus’ OKR journey began with just the six-person leadership team for the first two quarters of 2021. The organization had to pivot early on, Harris explained, because they were “​​too focused on revenue and sales, and it was tricky to get an engineering team to buy into how they can impact that.”

The top-level vision for the company, which produces a semi-bespoke software solution for enterprises, and includes a significant professional-services aspect, evolved to: “Revolutionize the way businesses get work done.”

During Q1, Lokulus underwent a company-wide restructuring and hired from the outside for a new engineering leadership role. This was an added challenge because the new teammate was not used to this type of goal-setting and felt two weeks into the role was an aggressively short time to suddenly have to hit such high targets.

As CEO, Harris said it’s his job to educate senior leadership: “This is your focus. This is what we’re striving to achieve. As long as you’re striving, the trust is there. If you don’t hit that, it’s about stretching.”

The Role of Psychological Safety in OKRs

As with all things in an organization, OKR success comes down to psychological safety. If your teams and individuals don’t feel safe to turn feedback into goals, they might not be ambitious enough. And if you’re putting too much top-down pressure on, they are likely to burn out.

It’s important to communicate continuously that OKRs are about reaching toward — not necessarily meeting — goals. That can be a big change from other ways of measuring performance.

And, as Harris put it, it’s “understanding if you fail to do your bit, it impacts the company.”

In Q3 2021, Lokulus brought OKRs down to the next level of senior management and team leads. Finally, this January, leadership rolled out individual OKRs to the whole company.

For Harris, getting cross-company buy-in comes down to full transparency, which includes updating his own OKRs during weekly town halls. This quarter, those OKRs are:

  • Individual Objective: To ensure Lokulus is a business that is data-driven in making decisions.
  • Key Result 1: 100% of staff are using the new OKR feedback model by the end of January, including everyone talking about their OKRs weekly.
  • Key Result 2: Ensure implantation of various toolsets, including customer relationship management, professional services automation, road mapping and reporting around existing customers.

As CEO, Harris’ focus is also on achieving company-wide OKRs:

  • Company Objective: Retain our existing clients and increase their lifetime value.
  • Key Result 1: Have a one-on-one meeting with all existing clients over the quarter.
  • Key Result 2: One-sheet account plan in place for all clients.
  • Key Result 3: Get a single customer view of all clients, mapping out their market.
  • Key Result 4: Complete product launch of new feature set.

Lokulus is only a year into OKRs. Harris suspects it’ll take a year or two more to get the full benefit of adopting OKRs. By then the company structure will probably evolve again to be less vertical and to create cross-functional, shared objectives. In the meantime, the company has regular off-sites and check-ins to make sure there’s cross-communication.

The next step is to move away from annual bonuses, Harris said. Some other incentives would be given quarterly, but in a way that “bonuses are not directly linked to OKRs and performance, but the whole concept of the short-term drive cycle.”

As with all things, open discussion and continuous improvement are the keys to OKR success.

Featured image by William Warby on Unsplash. Additional images courtesy of Rodríguez Pardo and Associates.