Target setting
A target is the number on a Key Result, and it has to do three things at once: stretch far enough to change how the work is approached, stay credible enough that the team believes it, and be fair enough that missing it by a little is survivable.
This is chapter 7 of From Trading to Scaling, “The Goals”. The book is free and takes about an hour.
The obvious bit
The number isn't the goal. It's a device for changing how people approach the work, and it only does that if they believe it.
Which is why the most common target-setting failure isn't the number being too high or too low. It's the number arriving without a reason, from a spreadsheet, and everyone quietly deciding it's decoration.
Where a defensible target comes from
The model, first. If you have a metric tree, the target is an argument: this lever moves that driver by roughly this much, so this is what a good cycle looks like. Without the model, you're picking a number and hoping.
Then the baseline. What happens if nobody does anything? Startling numbers of targets are quietly below trend, which is a target to look busy.
Then the stretch. Far enough that the current approach won't get there. That is the actual test of ambition, and it's a better test than a percentage: if the team can hit it by doing more of what they already do, it's a forecast, not a goal.
Write the reasoning down alongside the number. A target with its argument attached survives a change of personnel. A bare number does not.
Fairness, which isn't softness
Three conditions, and all three are structural rather than sentimental.
The resource arrived. A stretch target set without the capacity to reach it's not ambition, it's a message, and people read it correctly as decoration.
The lever is genuinely theirs. A target on something a team can't influence teaches learned helplessness in about one cycle.
An honest miss is survivable. If the team that reached hard and missed by ten percent has a worse quarter than the team that comfortably delivered a small number, you have set the real target, and it's lower than anything you wrote down.
The failure modes
The negotiated target. Two rounds of haggling produce a number both sides can live with and neither believes. Everyone learns that the first number is theatre.
The target that is also a bonus gate. Attach pay to the number and you haven't created ambition, you have created a strong incentive to set a low one. That is a design decision, and it should be made deliberately rather than by accident.
The precise target on an imprecise thing. A number to two decimal places on something you measure roughly is a confidence signal you haven't earned.
Sandbagging, treated as a personality problem. It's almost never one. It's a rational response to what the last few cycles rewarded.
Our opinion
Most companies argue about the number and ignore the two things that actually determine what people write down: whether the resource is real, and what happened last time to someone who reached and missed.
Get those two right and target setting becomes a straightforward conversation about the model. Get them wrong and no amount of encouragement to be ambitious will survive the first review.
One line to keep: a target people don't believe changes nothing, and a target that punishes an honest miss changes only what people are willing to write down.
How ambitious should a target be? +
Far enough that the current way of working won't reach it. That is a more useful test than any percentage, because it puts the emphasis on changing the approach rather than on the number itself.
Should targets be tied to bonuses? +
Only with your eyes open. Attaching pay to a target creates a strong, rational incentive to set a low one, so if you do it, design the gate deliberately rather than letting it happen by default.
Why do teams sandbag? +
Because something taught them to. Usually a review where an ambitious miss went worse than a comfortable hit, or a stretch goal that arrived with no extra resource. Sandbagging is a designed behaviour, not a trait.
From the ZOKRI OKR Handbook, the methodology we install and maintain. Written by Matt Roberts.
Matt leads enterprise OKR implementations at ZOKRI with Peter Kerr. Across the team that's 100+ implementations in 15+ enterprise organisations, from banks to car makers, in several countries and rather more time zones. Before ZOKRI he co-founded Linkdex, an enterprise SaaS platform, and sold it. He writes the method behind these notes, most of it learned the hard way in other people's leadership meetings.
Target setting is where most of the honesty in a goal system is won or lost. We bring the model, the questions and the awkward comparison to last year.
Already running OKRs? Get an objective read on the implementation: the OKR Implementation Review, free, 45 minutes.