March 24, 2026
Hi Everyone,
If you're setting Q2 targets right now, there's a good chance you're defaulting to stretch goals because they feel ambitious and leadership-worthy.
But about 80% of participants in an Organization Science study failed to reach their stretch goals. The stretch targets also increased risk-taking and led many people to quietly abandon the goal in favor of a lower, self-set one. The handful that succeeded were the exceptions. For most, stretch goals worsened performance.
Today, we're walking you through a method for setting targets that balances ambition with reality.
Every metric needs a range
The problem with most target-setting is that each metric gets a single number. You hit it or you miss it.
That binary nature creates two bad situations:
- If the target is too easy, people coast.
- If it's too aggressive, they disengage or take shortcuts.
One fix that works well is to set 3 numbers for each important metric.
Threshold is the minimum acceptable level. If you fall below this, something is genuinely wrong and needs immediate attention. Think of it as your early warning line. A common rule of thumb is to set the threshold at a level you'd expect to hit about 80% of the time.
Target is your realistic expected performance, the number you're planning around and staffing toward. It should be grounded in what your team has delivered recently, adjusted for any known changes in conditions. This is the number that goes into your operating plan and resource allocation.
Stretch is your aspirational level, the result you'd get if several things went your way and execution is sharp. This is the number that requires new thinking, not just more effort. Set it at a level you'd expect to achieve roughly 20% of the time.
Using all three gives your team a realistic goal to commit to while preserving upside ambition. Performance conversations get more useful too, because you're asking "where did we land across the three cases, and what drove it?" instead of just "did we hit the number?"
Pair every number with a quality check
This is the piece that trips up a lot of teams. When you set a target for one metric, people will find ways to hit it, and some of those ways will hurt you.
For example:
- Call centers that target average handle time get agents who rush customers off the phone.
- Developers measured by lines of code write bloated, harder-to-maintain software.
- Sales teams measured on deals closed will discount heavily to pull revenue forward.
- Support teams measured on tickets resolved will close cases before the customer's problem is actually fixed.
The economist Charles Goodhart put it simply: when a measure becomes a target, it stops being a good measure.
So for every quantity metric, add a quality metric that would expose gaming. If you're tracking deals closed, also track the average deal margin. If you're measuring response time, also measure first-contact resolution.
The pairing makes it very hard to game one number without the other one showing the damage.
Amazon runs this as a formal practice – every metric in their system is paired with a counter-metric designed to catch exactly this kind of distortion. If someone games the speed metric, the quality metric will show it.
Go deeper
👉 KPI Tree: Goodhart's Law and Metric Design – a guide to pairing quantity metrics with quality counters, with worked examples across industries
👉 Stephen Lynch: KPI Best Practices from Amazon – how Amazon manages metric lifecycles, pairs input metrics with quality guards, and runs weekly business reviews
👉 Equity Methods: Best Practices in Performance Goal-Setting – a walkthrough of threshold/target/stretch calibration using historical data and peer benchmarks
👉 BCG: The Art of Performance Management – how to identify the KPIs that matter and build a system that turns data into decisions
Coming up tomorrow
Tomorrow we're sharing a one-hour exercise that shows you exactly how to fix your pricing and who's responsible.
Thanks for reading!
P.S. Does your team use stretch goals? Let us know, we’re curious whether they work for you.