Training Your OKR Muscle: Habits, Metrics, Gotchas

I tracked OKRs in my personal life to prepare myself for the challenge of using them at work.

I knew I would be missing the core components of the process - cross-org alignment, team buy-in - but getting the habit right was more important than those specifics, which vary org to org anyway.

Each day, I logged my progress and had to look for ways to progress.

I learned that:

• It is hard to come up with the right Key Results & KPIs even if you know the physics of the underlying process.
• It's easy to start gaming whatever metric you pick even without a monetary incentive attached to hitting it.
• I had to rethink my Key Results several times during the period, based on new information. I didn't shift the goal post as much as adjusted the measurement itself.

Don't do it in a spreadsheet

I used a tool called Tability to manage OKRs (I wouldn't recommend it though - it has bugs).

Tability & other OKR software are mostly glorified spreadsheets with reminders, but I wanted to save myself time to create this stuff from scratch.

In this case, I wasn't worried about lo-fi prototyping the solution, I was worried about forgetting the myriad of tiny gotchas in the OKR process.

Don't do it in spreadsheets unless you really know what you need to track.

I focused on work objectives.

Example & mistake

Work objectives were about hitting the first successful 30 days on the job.

What they looked like:

My biggest mistake: not focusing on the user.

We know to focus on outcomes, not outputs. We know to use numbers and dates vs. binary yes/no outcomes.

I did my best to control for that, but even if KR's are outcomes, it may still vary by who perceives these outcomes. In my case, I could feel like I had "successful first 30 days" by the end as expressed by hitting the KRs, but that perception could be different for my team members, peers & managers.

The key here is that the consumer of your efforts (i.e., the person whose behavior you’re trying to change) is a third party — not yourself. - HBR

The Good Stuff

There were good things I'll stand by...

• I checked in daily vs. weekly. While it could be tough to measure progress daily in your team, weekly is just not often enough to set a habit. And while I could still be doing daily things, but checking in weekly, the chance of missing the weekly check-in would be higher. Weekly check-in creates a lag, and you want faster feedback here.
• Low number of OKRs. Right away, I knew that less was more. I kept having to remind myself that besides counting objectives, I have to count the Key Results under them. Saying you have 2 objectives with 3 key results each means you have 6 pieces of data to track. That's already a lot.

Insights From Another Domain

Starting a new role is stressful, so I had to prepare to make sure I don't fall off the bandwagon. So I ran Health OKRs in parallel.

Here's where I learned something much deeper.

Suppose we take fat loss as one main objective. It's a well-understood space with years of research, and what's more important, I knew the research.

I knew that in order to lose fat, you needed to be in a caloric deficit, and your muscle mass would contribute to this equation, so you needed to grow that as well. I knew that growing muscle mass would depend on using progressive overload & increasing total volume (weight * sets * reps). My KPIs were in a pretty good shape.

Insight #1

To have actionable clarity in your OKRs, you need to know the physics of the system in question well.

I relied on years of research by others, and on years of personal curiosity about this space. What if in your line of work, everything is new and unknown?

It doesn't mean you have to wait years until someone discovers your arcane niche and builds a body of research.

Things to try:

• Find a space that's like your space and pull the physics from there. One of our team's tackling adoption for an internal tool. Other people have done this in other domains and have much to teach. So use that.
• Do discovery yourself, make room for it in your work, but just enough. No one set out to run the One Big Experiment to determine the best pathways to losing weight. It was comprised of many small pieces of research. But the point is: they also made room for that research instead of pretending that they know it all, ploughing ahead without results.

Insight #2

Metrics will be gamed even if the incentive is just to hit the goal.

For my "Total Volume" metric, when I couldn't hit a weekly growth target, I started to do exercises where I could do more volume (and thus hit the number) without breaking myself.

You can't do 5 sets of 25 reps of 100 kgs on your bench press unless you're an athlete. But on a leg press, that's much easier to do. So I did this exercise more instead.

But that meant I could skip a day of workouts while doing easier workouts.

I was hitting the number, but not affecting what it was meant to represent.

Things to try:

• Add nuance metrics to balance. North Star metrics are oversimplified by first-timers. The true North Star approach is more nuanced, and you need to balance it properly to avoid the traps of local optimisation & gaming the metric. In my case, I framed it by forcing not to do the same exercise more than once a week. I depleted my cheat exercises and was forced to do it right (until I would try to game this system again).
• Relax your criteria. For the core metric, make sure that there are other unrelated metrics that signal progress to a goal. If your Key Results are all around the same metric, you will have nothing else to do but to try to hit the target. In my case, I had non-workout Key Results that were more around quality of life, eating habits & variations in my workouts.

--

Hope you found this useful. I'm still learning to use this tool. It's not the be-all and end-all, but it was fun to explore.