Set the Right Stretch for Team Goals Without Burning Out

The change that helped most was stopping the habit of raising targets after an early win. Many teams reward strong early performance by moving the finish line higher. That often leads people to hold back on forecasts or save their effort for later. We chose a different approach by keeping the original goal and using any extra progress for a stretch objective selected by the team.

That approach built trust while still encouraging people to aim higher. Everyone knew their success would be recognized, which increased effort from the beginning of each cycle. Teams also made better decisions because they were not trying to reach higher targets after plans had already been set. Performance improved without lowering morale because the goals stayed fair and ambition felt natural.

I pull last cycle's numbers and find the goal my team hit at a partial completion rate. That becomes my baseline, and I set the target a stretch above it. If they completed most of a goal, they had enough momentum to push further.
The change that improved performance was breaking the cycle target into shorter checkpoints. We used to set a quarterly goal and check progress at the halfway mark, and if we were behind by then, the energy was already gone. Now I split the quarter into smaller windows so we can see traction building in real time, and I can adjust the difficulty up or down before anyone gets demoralized or coasts. My team started finishing cycles stronger, and I keep the ambition high because the feedback loop is tight enough that nobody is guessing where they stand.

For a content team hitting a new quarter, I used to set targets based on what I wanted the number to be, not what the team's actual recent output supported, and it backfired. I once set a target of 20 published articles for the quarter based on a good month we had, and the team hit 11, then started quietly missing weekly check-ins because the gap felt too big to close.
The change I made: I now set the target at roughly 15% above the trailing three-month average, not above our best single month. For that same content team, trailing average was around 13 articles a quarter, so the new target became 15, a stretch but a visibly reachable one.
They hit 16. Past that, I've noticed the team pushes itself harder toward a target that's slightly above what they know they can already do than toward one that's obviously aspirational, because the close ones feel winnable and the distant ones just get quietly written off after the first bad week.

I used to set goals that made my team feel like heroes when they hit them. That was the problem.

When I was scaling my fulfillment company toward that $10M run rate, I'd announce quarterly targets in all-hands meetings and watch everyone nod enthusiastically. We'd hit maybe 80% of them. I thought that was winning. Then my COO pulled me aside and said something that changed everything: "Joe, we're celebrating mediocrity. Your goals feel like stretch goals, but they're actually just math on our current trajectory."

He was right. I was taking our monthly growth rate, extrapolating it, adding 15%, and calling it ambitious. Real ambition meant asking what would break if we tried to double that number.

Here's the specific change I made: I started setting two goal lines instead of one. The first was "expected performance" based on current capacity and systems. That's what we budgeted around and what determined bonuses. The second was "breakthrough performance" that required us to fundamentally change how we operated. Hit the first line, you get paid and celebrated. Hit the second, you get equity or a promotion conversation.

For warehouse productivity, expected was 150 orders per person per shift. Breakthrough was 200. Hitting 150 meant we were executing well. Hitting 200 meant someone figured out a better pick path, negotiated new equipment, or redesigned our receiving process. It separated "working hard" from "working differently."

What shocked me was how many people aimed for breakthrough once they understood it wasn't about working more hours. One of our shift leads redesigned our packing stations and got us to 190 orders per person in six weeks. I promoted her to operations manager.

The mistake most founders make is setting one ambitious number and hoping everyone rises to it. That just creates anxiety. Give people a floor they can confidently reach and a ceiling that requires invention. Performance goes up because suddenly the ambitious goal isn't about effort, it's about creativity.

I choose a target by asking one simple question. What goal would make the team rethink their current habits without making them doubt the plan? If a goal can be reached by doing the same work a little faster, it is too safe. The right target sits in the middle where better judgment, clear priorities, and stronger teamwork become necessary.

One change that helped was adding a review before setting the goal that focused on real constraints instead of optimism. We looked at time, skills, dependencies, and hidden challenges before deciding on the goal. That kept ambition from being based on assumptions and made planning more realistic. The result was better performance because teams spent less time fixing problems and felt leadership respected the reality of getting the work done.

What makes a target feel possible? Not its size. The teams I set these for run 6 to 9 people, small enough that everyone can see whether a number is honest or decorative. We used to pick the stretch figure first and argue backwards. Now the number starts at the best month the team has actually had and moves up by whatever they can name a reason for. If nobody in the room can say what would have to be different, it comes back down. That conversation takes 20 minutes and it is the whole exercise.
Effort went up when the target got smaller, which I did not expect. You stop hearing people quietly decide in week 2 that the quarter is already gone. At 40 people, where whoever sets the number has not watched anyone work, I would not bet on it holding.

When I set team goals I start by creating shared, practical context so everyone understands what success looks like and why the target matters. I use in-person or structured virtual sessions to walk through examples and to surface questions that reveal whether a target is realistic. The single change that helped calibrate ambition was shifting from passive communication to direct education and prioritizing face-to-face meetings. That change improved alignment and reduced confusion, which raised performance without hurting morale.

The goal-setting mistake I made most consistently early at Tibicle was confusing ambitious with unclear. I would set targets that sounded challenging but had no specific measure attached to them. Ship more. Improve quality. Communicate better. The team nodded and nothing changed because nobody knew exactly what success looked like.

The shift that raised performance without damaging morale was making every target binary. Either the sprint delivered the defined features by Friday or it did not. Either client response time stayed within the agreed window or it did not. No partial credit, no interpretation required.

Binary targets feel harsh until you calibrate them correctly. The calibration step that made the difference was involving the team in setting the number rather than handing it down. A developer who agrees that four tasks per sprint is realistic will work toward four. A developer who had five imposed from above starts the sprint already feeling behind.

The unexpected outcome was that team-set targets were consistently more ambitious than what I would have set for them. People who understand their own capacity and feel trusted to use it honestly set harder goals than managers who are trying to be encouraging without full information.

Ambitious targets land well when the person hitting them helped define them.

The change that mattered was separating what the team controls from what they merely influence, and scoring only the first.

We used to set targets on client outcomes. Positions, enquiries, revenue. It sounds accountable and it is corrosive, because search and paid results move on a lag and for reasons nobody in the room caused. A team can do the best work of the year and watch the number go the wrong way because an update landed or a competitor doubled their budget.

When we looked back across our own accounts, about 76% of the month to month movement traced either to work done in the previous quarter or to something entirely outside our hands. Scoring people against that is scoring them on the weather.

So the committed goal is now built from inputs. The volume and standard of work shipped, the tests run, the technical debt cleared. Outcomes are still tracked in the open, still argued about, still the thing that decides what we do next. They are not what anybody is judged on.

The calibration test I apply is whether somebody could hit their goal while doing poor work. If they could, the goal is written badly. The moment it landed for me was a quarter where the team hit every output target and I still had to tell them results were flat. That conversation was calm and useful, which it had never been while the outcome was the verdict.

The change I made was pairing every ambition number with a guardrail number, and treating the guardrail as the harder commitment.

Ambitious targets are cheap to set and easy to hit badly. You can grow signups by loosening who you sell to. You can grow revenue by taking on accounts you will spend a year regretting. The number goes up, the company gets worse, and nobody in the room has lied about anything. That is what damages morale, not the height of the target. People can tell when a win was bought with something.

So each cycle we set the growth number and, next to it, a line we will not cross to reach it. Monthly churn stays under 2%. Support replies stay inside the same day. If hitting the target would require crossing the line, the target was wrong, and we say that out loud instead of quietly choosing one.

It settles the calibration question by itself. Once you cannot buy the number with damage, ambition has to come out of the work, and the honest ceiling on the work is obvious to everybody doing it.

The quarter that taught me this was one where we hit a growth figure I had been pushing hard for and churn moved right along with it. Same revenue on paper, worse company. I had asked for one number and gotten exactly that.

Set the floor before the ceiling. The floor is what people trust you on.

I look for a number the team can picture the path to, even if they can't yet picture finishing. If nobody can describe a plausible first month, the goal is a wish and people quietly stop believing you.
Years ago we ran a weightlifting fundraiser at Penn State and raised about $10,000 the first year. The second year we raised $50,000. That jump wasn't a spreadsheet exercise. It came from asking what number would make the effort clearly worth it to the people showing up, then working backward to see whether we had enough hands to get there. Ambition set that way pulls people forward. Ambition set to look good in a board deck pushes them away.
The calibration change that mattered most for me was saying out loud, before the cycle starts, what happens if we miss. Usually the honest answer is that we learn something and adjust. Once people know missing won't cost them their standing, they'll accept a harder target. I learned that in sports. Relieve the fear of failure and people play free, and they play better than they will while they're protecting themselves.

Over my ten years of managing engineering teams and scaling AI platforms, from my time at Leboncoin to building our customer agent OS at AGO, balancing ambition and realism usually comes down to separating predictable work from unpredictable research.
In the past, we tried to set single, massive targets for a development cycle, like launching a new autonomous support feature and mandating it hit a 95% success rate out of the gate. That usually hurt morale. Hitting a strict accuracy threshold with language models involves trial, error, and research that doesn't always fit neatly into a standard sprint cycle.
The change we made to calibrate our ambition without burning out the team was splitting our cycle goals into deterministic and probabilistic buckets. For deterministic work—like building an API connector or updating our knowledge base orchestration pipeline—we set aggressive, tight deadlines because the technical path is entirely known. We know exactly how much effort it takes, so we can set a target that stretches the team's velocity just enough.
But for probabilistic work—like tuning an agent to handle edge-case customer frustrations without hallucinating—we set goals around a time-boxed research effort and a baseline improvement, rather than demanding a hard, absolute accuracy metric that might not even be feasible. Structuring our cycles this way kept our release pipeline moving and raised overall performance, largely because the engineering team stopped feeling paralyzed by being graded on the inherent unpredictability of AI.

In marketing, it's unfortunately easy to set aggressive growth targets without accounting for changes in the budget, hiring, or customer behavior. One thing that I have learned over my time leading marketing teams is that past performance isn't always a useful standard to apply for setting current goals. We have moved away from using last year's numbers to set our targets, and instead now build them around the team's current capacity, market conditions, and the specific initiatives we know will drive results. This also helps to craft goals that allow team members to clearly see the logic behind them, which as a rule makes them far more motivating.

A practical framework that we use is to establish a baseline expectation along with a target goal and a stretch goal. We make the target ambitious enough that it requires new ideas and better execution, but still achievable with strong performance. The stretch goal represents exceptional performance. This shift improved both performance and morale. Teams no longer felt that they had failed if they missed an overly aggressive number, and they could see progress against meaningful milestones throughout the quarter and stay focused on continuous improvement. That flexibility also allowed us to adjust priorities without making people feel like the finish line is constantly moving.

I stopped measuring productivity the way most teams do. When you only have three people shipping five product lines, traditional goal-setting becomes a trap. Setting a goal like "ship prediction markets by end of Q2" sounds concrete, but it ignores what actually slows you down: context-switching, architectural debt, and the coordination overhead that multiplies every time you add a dependency between people.

The change I made was to frame cycle goals around what I call "execution surface reduction." Instead of setting feature-count targets, I started defining what we were explicitly not going to touch for the next six weeks. That sounds like it would lower ambition, but it had the opposite effect. By deciding up front that we weren't going to revisit the staking interface, weren't going to add another chain until the existing plumbing stabilized, and weren't going to prototype new AI features until NikaAI shipped to production, we created a forcing function for depth over breadth.

The team shipped faster because every decision that came up during the cycle already had a default answer: not in scope. We routed perpetuals through Hyperliquid via builder codes and prediction markets through Polymarket instead of building matching engines or oracle stacks in-house, which kept our internal engineering surface narrow even as the product surface expanded. That routing pattern is what makes a three-person team competitive with organizations ten times our size.

Output went up because friction went down. Morale stayed high because people could see real momentum on the things we were actually touching, rather than feeling like we were maintaining ten half-finished surfaces at once. The calibration shift was simple: stop defining goals by what you want to add, and start defining them by what you're willing to ignore.