Debugging Your OKRs

January 11, 2022

Love ‘em or hate ‘em, OKRs (Objectives and Key Results) are practically ubiquitous in the world of startups.

Even if your company doesn’t actively do OKRs today, it’s likely you work alongside somebody who’s had some experience with OKRs in the past.

Before joining Netlify, I worked for several years at a company that was so exceptionally awesome at running OKRs, the whole practice of it become permanently imprinted in the way I work. I can’t shake the feeling of OKRs, even when they’re not happening in my workplace.

Now, not everybody at my last org was as gung-ho about the OKR cycle as I was. Fair criticism was voiced. Leaders lamented the logistical overhead. New managers resisted the bottoms-up planning.

And certainly as that organization grew from 50 to 200+ people, the original design of that company’s OKR cycle had to bend and stretch. We were always experimenting with different lenses of looking at the process, tweaking the rules here and there.

At this point, it’s been so long since I’ve read Measure What Matters I can’t even recall all the tenents of the officially blessed way to run OKRs.

But, what I do know is how I personally like the OKRs to work, how I remember them best, and what I recognize as red flags or indicators of OKRS gone awry.

Let’s dig in!

12 OKR Red Flags

1. Too many objectives.

Things start to break down when there are more than 4 objectives at the company-level. Between 2-3 objectives is a good sign that the business is properly focused.

2. Too many objectives with explicit numbers.

It’s okay for some objectives to include a specific number.

For example: Hit more than $10M ARR.

However, if all of your objectives include numbers, this feels off to me. In my mind, the perfect objective paints a vision of where you want to be at the end of an OKR cycle. It should feel lofty, aspirational, and transformative.

Leave the tactical numbers for the Key Results (KRs).

3. KRs look like to-do list tasks instead of results.

Key Results should be specific and clearly measurable without being prescriptive about how you achieve the results.

This is where you have lots of number targets, like:

  • Increase signups from best-fit customers by 50%
  • Increase feature adoption of x feature from x to y
  • Bump our email click-through-rate to above 10% at least 3 times

Having KRs that are clearly tied to numbers gives the team maximum flexibility to experiment with projects and tactics to hit the numbers.

4. Only leaders and managers own KRs.

Company leaders should be in charge of setting objectives. But, it’s an anti-pattern for department heads and people managers to always own every KR on the board.

Especially as the company grows and scales, the most senior leaders should become sponsors or contributors for KRs without always driving or owning the numbers themselves.

When OKRs are done right, they give skilled contributors and informal leaders a chance to identify the efforts that will most likely help the company achieve the desired outcomes, to bubble up the great opportunities that the senior leaders may not see from where they’re seated.

5. Every single person in the company owns a KR.

If too many people have OKRs on the board, the likelihood for un-foreseen dependencies skyrockets.

Or, individuals end up scoping their OKRs to work within their own individual control, which turns the whole OKR board into a punchlist of personal to-do items.

Ideally, every person in the company can identify at least one KR on the board where their work is contributing or having an impact. But, it’s not necessary for every single person to own a KR every cycle.

In a healthy org, individuals on teams may effectively take turns, stepping into the reporting role or the contributing role, from cycle to cycle.

Which reminds me of another red flag.

6. OKRs become tied to leveling and promotions.

If the criteria for getting promoted from a Level 3 to a Level 4 is “almost always leads reporting of a KR” then guess what—you’re going to have ICs fighting over who gets to lead the KRs.

Don’t do this.

7. Too many perfect scores.

OKRs should push the company to achieve major stretch goals. Overall, when you average your scores together, you should be looking for a 0.7 across the whole org.

If everybody is getting a 1.0 perfect score every time, they’re sandbagging it.

8. Too many scoring rubrics relying on pass/fail.

Similar to the problem where OKRs devolve into a public-facing task list. If all of the scoring rubrics are 0.0 or 1.0 pass fail, with no incremental wins in between, this is a bad sign.

Pass / fail scores should be reserved for high-stakes mission critical work that truly doesn’t have any meaningful milestones along the way.

9. The KR scoring is tied to a metric—but there’s no baseline for the prior cycle.

I’m a big believer in the “graceful zero” which is when an OKR owner identifies early on that the KR is doomed and shuts the whole thing down rather than spin company resources trying to force some points on the board. Take that energy and put it elsewhere!

But, there’s another type of “zero” in OKRs which is less graceful and more common: the unscorable KR because a critical metric is impossible to report.

If your KR is tied to a metric that your team can’t measure yet, you’re in trouble. Measuring new metrics is hard work. Things have to be instrumented properly. Data has to be both flowing into the right systems and reliably accessible.

Don’t tie your KR to a metric you haven’t measured for at least a few weeks in advance.

If the perfect metric doesn’t exist yet, pick a reliable proxy.

Otherwise, build the work of reporting the metric before a specific date into the OKR scoring rubric, i.e.

  • 0.5 points for reporting this metric before x date in the cycle

10. OKRs happen in isolation.

I’m totally guilty of this one.

Last year I tried to roll out OKRs only for a specific initiative I was leading, because it’s the way I personally like to work and set big goals.

Pretty sure I confused the crap out of people and frustrated my team. I lost lots of valuable time asking questions like, “Yea, but what’s our desired outcome?”

If you’re going to do OKRs, everybody has to be bought in. It needs objective-level sponsorship and some shared understanding.

Forcing OKRs onto a specific team or project when people aren’t bought into the concept is risky business.

11. Every department has their own KRs and there’s no cross-pollination.

OKRs work best when you’re inviting contributors from multiple teams and disparate parts of the org to solve the same big problems together.

If every department head treats their KRs like a list of their own team’s goals, you’re likely to get duplication of efforts, where various teams across the org are trying to solve the same problem but not collaborating.

This defeats the whole purpose.

12. OKRs fail to cascade down.

When objectives get set at the company-level, teams should be empowered to react to the objectives and start proposing the right KRs to measure.

By the time the KRs are locked in and baked, people should have a sense of what projects, activities, and experiments will “roll up” to the OKRs.

If the objectives are presented and the KRs get locked in but everybody shows up with execution plans that look like business-as-usual, this was a bust.

Great OKRs should inspire people to adjust their plans. Say “no” and carve away busy work that doesn’t support the shared goals. Think big to imagine high-impact projects to ratchet the business forward.

With all of these ways to do OKRs wrong, it’s understandable why some people have an unsavory experience with them.

I still think they’re fabulous.

Looking for more content on OKRs? Check out this older post where I break down the whole process and share a template for OKR reporting.