How to score OKRs properly

Some people score themselves too harshly (not giving themselves enough credit), and some people score their OKRs too softly (giving themselves way too much credit!).

If you are caught cheating on your scores, you will lose trust with your stakeholders. And if you under-value the work the team is doing, you'll lose their support.

But you've got to be honest about what the scores are, even if they're not as high as you'd like, to gain everyone's respect.

Different types of Key Result

There are two types of key result that you can have in your OKRs and they are scored differently.

The first is a Committed key result. These KRs are either "done", or "not done". There is no in-between. These are usually deliverables or outputs or deadlines that you're trying to meet.

The second is an Aspirational key result. These KRs are usually a metric you're trying to move or a target you're trying to hit. So you can score partial progress towards that goal, or even over-achievement if you manage to exceed it.

Let's dive into the specifics of each type, and you can go here to learn more about the different types of OKR.

How to score Committed key results

A committed key result is either "done", or "not done". So the scoring is binary.

If the KR is not met, you have to score 0.0. That means if you had a delivery deadline and you missed it, 0.0. Or if you had a deliverable you had to ship and it either didn't get shipped during the quarter, or it didn't get successfully shipped (eg. actually in use by customers without any major bugs), then you score 0.0. Or if you only managed to ship part of it, that's a 0.0 too.

If the KR is met, you should score 1.0. If you did what you committed to in the KR, you get full marks.

Obviously there are still grey areas though...

Sometimes during the quarter you might learn new things that affect your decision about the KR.

For example, perhaps you find out that the effort required to build the feature is actually way bigger than initially estimated, and so you cut scope or change the solution to fit it into the quarter and customers still get the benefit you were trying to deliver. In which case you could argue it's a 1.0 score. But it depends on how you phrase the KR!

If you said: "Deliver feature X" then you might have to score 0.0. But if you said "Customers can achieve Y" then you could still score a 1.0 even if you changed the solution.

Or perhaps you learn that the feature you planned to build is not actually useful. In which case it's OK to not deliver and score a 0.0 and be happy with it. If you ended up building something different then you could add a KR to cover that, and score yourself 1.0 instead (as long as you achieved it!).

Another example could be if you had a delivery date Release X to production on 01/05/23" and you were 1-2 days over but this was due to unexpected sickness in the team, you failed the target and score 0.0. If you had a KR that was phrased as "Spend 20 days on feature X" then you could count only the days when people were actually working on it and exclude sick days, and still give yourself a 1.0.

Committed KRs should be used sparingly, because the need to get a 1.0 score causes teams to try and game the KR itself (as above) to ensure they hit it. Instead, try to keep all your KRs as aspirational.

How to score Aspirational key results

Scoring aspirational KRs can be subjective. You have to decide how much progress you've made according to the scale below. We've included examples of situations when you might choose to give each score.

0.0 - 0.3 (red)

• little to no progress made
• nothing shipped
• if you had a metric target, it didn't move
• or the metric moved but is a long way away from the target

0.4 - 0.6 (amber)

• some progress
• some stuff shipped, perhaps partial scope or reduced functionality
• metric moved, but not enough or in the wrong direction (at least you learned!)
• done your bit but blocked by support needed from other teams

0.7 (green)

• target achieved
• metric moved all the way to hit the goal that you set
• everything shipped, or enough shipped to hit the goal

0.8 - 1.0 (green or blue)

• target exceeded
• the metric moved further than hoped
• you shipped more stuff
• you shipped faster, with less effort than expected
• extra work happened elsewhere in the company that was a bonus

In some cases you can treat the score as a percentage. For example, if your KR is:

Increase conversion by 1% from 1.5% to 2.5%

And you get it to 2.0%, you could say you're 50% of the way towards the target and give yourself a 0.5 score. This would be reasonable - you shipped some stuff, the metric moved in the right direction, but didn't hit the target.

But that approach doesn't always work. Imagine your KR was:

Achieve 5% conversion with at least 3 new marketing campaigns

And you get 3% conversion on the first, 4% on the second and 6% on the third - what's the overall percentage then? It can't be calculated so easily. In this case you're still in the amber zone: you shipped some stuff and you had some successes but didn't reach the target. Depending on how strong your results are you can choose between 0.4 and 0.6. With the results mentioned here, a 0.5 would be about right.

If you got 6%, 7% and 4.5% you would definitely qualify for a 0.6 score because you're so close! Maybe even higher since you had multiple campaigns over the target of 5%. If you learned a great deal about how to break that threshold and you strongly believe it's repeatable, you could give yourself a 0.7.

Can I change key results mid-quarter to hit the target?

Yes... and no.

OKRs are not set in stone, they can always be iterated and tweaked. But you've got to do it genuinely and not because you're trying to cheat on your scores.

When you need to change something about your OKRs, it's a pivot.

For small pivots you can just change the KR, inform your stakeholders of the change and rationale, and carry on:

• You learned about the opportunity size so your targets have to change
• You learned about customer behaviour, so the metric has to change
• A competitor is also working on something similar, so your delivery deadline has to change

For big pivots you should keep the original KR but deprecate it somehow (cross it out, colour it grey, put a "(DESCOPED)" label on it, whatever), and add a new one. Again, inform your stakeholders of the change and the rationale, and carry on:

• A law or regulation changed and the feature is no longer possible, so the KR must be scrapped
• A technology spike proved that the effort required is too large, so perhaps the O needs to be changed
• User research shows that the feature won't solve the customer problem, so the KR must be different

You get the idea. It's OK to change your OKRs, but only on the back of new information that changes the context. Not just because you're falling behind the target.

That's all for now.

To score your OKRs properly you must be honest and open with yourself, your team and your stakeholders. This will earn respect, build trust and garner support. As the saying goes, cheaters never win, and winners never cheat.

P.S. Use an OKR tracker app to stay on top of your progress and display your scores visually in a dashboard. It's free!