Stretch Goal

Goal-setting sessions turning into negotiations is an old problem. When hitting the number feeds straight into a review and a bonus, whoever writes the number picks one they're confident of clearing. That isn't laziness; it's the rational move the system rewards. Ask ten managers for a target that will decide their own rating and ten conservative numbers come back.

A quarterly planning meeting looks like this. Paid teams currently sit at 40 and the team lead proposes 50. Privately they think 60 is reachable on a good quarter, and the number they write down is 50. The quarter ends at 52, gets recorded as 104 percent, and nobody is worse off. The company is the only party out the eight it didn't go get.

The opposite failure is just as common: setting the number high and then grading against that height. Write 100, finish at 72, and have 72 percent carried into a review, and that team writes 55 next quarter. Nobody who took a loss once takes the same bet twice.

A stretch goal is the device for handling both at once. It raises the height and, in the same motion, removes the penalty for falling short. Doing only one half doesn't work. Raise the height while keeping the penalty and people write smaller numbers; remove the penalty without raising the height and the goals go slack.

Calculate what the current approach reaches, write a number visibly above it, and agree on a lowered completion bar before anything gets committed. Five steps.

1. Work out the baseline — where you land by keeping the current approach and improving it a bit. Starting from 40 paid teams, the baseline is around 50.
2. Set the goal somewhere between 1.5 and 2 times the baseline. In this example, 100. Go to something like ten times and people stop doing the arithmetic and stop looking at the number.
3. Agree on the completion bar before the goal is written. Get the sentence saying 70 percent counts as strong into the document first. Lowering the bar after the quarter closes does not produce the same effect, because nobody believes it the next quarter.
4. Keep this goal's attainment out of review and bonus calculations. Put it in and next quarter's number comes down.
5. At quarter end, write down what was learned instead of assigning a score. Which of the things tried on the way to 72 worked is the material for the next quarter.

The reason the device works is that the height changes the approach. Fifty is reachable by doing the current work a little better. One hundred is not, so it forces looking for a different method. The question shifts from how much harder to work into which channel nobody has tried yet. When Pagedeck aimed from 40 at 100 and finished at 72, aiming at 50 would have stopped it around 52. This is the place where a goal you missed leaves better results than one you hit.

Not every goal gets set this way. Anything where falling short is itself the problem — a delivery date promised to a customer, a regulatory requirement — runs at 100 percent, and the phrase 70 percent has no meaning there.

It's the default character of a goal in OKR. The key results attached to an objective get written at values that are hard to reach, and scoring 0.7 at quarter end is commonly treated as enough.

It does not get attached to a key performance indicator (KPI). Monthly churn or response time are numbers kept permanently on to catch things getting worse, so what's needed is a line not to cross rather than a height to aim at. Make response time a stretch goal where 70 percent compliance is fine and the indicator stops meaning anything.

It suits work where the baseline itself is uncertain. Entering a new market, or shipping something that didn't exist, leaves no grounds for calculating where the current approach lands. With no grounds to go low either, setting it high and watching what works is the better bet. A team launching in a country it has never sold in has no prior quarter to improve on, so any number is a guess and a high one at least provokes more attempts.

One condition has to be in place: people being able to say they fell short, which is psychological safety. Where that's low, the quarter-end report reads 98 rather than 72, and the next plan gets built on a number that was edited. Cascading goals need care for the same reason. A stretch number is fine at team level, but push it down into individual goals and it touches someone's review directly.

"Let's make this a stretch. Seventy percent is a good quarter."

Raising the height while setting the completion bar lower in advance.

"The target is 100 but it doesn't go into reviews. That was agreed up front."

Reconfirming that missing it carries no penalty.

"Response time isn't a stretch target. That one needs a line."

Marking which indicators this approach doesn't apply to.

Pagedeck wrote paid teams from 40 to 100 last quarter. The baseline Ravi calculated was 50 — where improving the existing ads and in-product prompts would land.

After writing 100, what Ravi brought was a different method. Noticing that inviting a teammate into the same document turns that whole team into a conversion target, he redirected the work toward the invitation flow. The change shipped in the second month, which left six weeks to see what it did. The quarter closed at 72. What Grace took from the retrospective wasn't the 72 percent; it was that the invitation path ran about three times more efficiently than the ads.

The most common reading is that higher is always better. Push the explanation that a high number forces a different method and you arrive there. But around ten times the baseline, people stop calculating, because a number nothing can reach gives no reason to change the approach. Left as is, the goal document becomes a declaration nobody reads, and the actual plan gets written separately next to it. The number on the page and the number the team is working toward stop being the same thing, which is a worse position than having set a modest goal.

The other reading treats falling short as failure. The ordinary understanding that goals are things you hit makes that automatic. But this approach only holds where the completion bar was lowered by agreement. Writing 100 and reaching 72 is a better outcome than writing 50 and reaching 52, and processing the first as a miss produces a team that writes 55 next quarter. Left as is, the mechanism stays on the page while the numbers in it get smaller every year.