If No One Can Hit the Number Honestly, Someone Will Hit It Anyway
By Derek Neighbors on October 7, 2026
Every planning season someone puts an unrealistic goal on a slide that nobody in the room believes: ship the platform rewrite by March, cut incidents in half this quarter, have agents writing half our code by June. The leader feels better as soon as the number goes up, because a big number feels like motion. The engineers do the math in their heads and get quiet.
Then March comes and the number gets hit anyway. The rewrite ships with half its features behind a flag nobody plans to turn on. The incident count drops because three outages got filed as “degradations.” Somebody hit the number, and nobody can explain how in a sentence they would want their boss to read.
I have been in that planning meeting more times than I can count, on both sides of the slide. The pattern is reliable enough that I now treat it as a law of organizations. Most people do not cheat, and somebody usually does. If no one can hit the number honestly, someone will hit it anyway.
Two goals that look the same on a slide
Put two goals side by side. Both are big, and both get applause at the all-hands. From the back of the room you cannot tell them apart.
The difference is whether someone close to the work can describe a route to the number, out loud, in a way they would be comfortable explaining to the person who set it. The route can be uncertain, but it has to exist. One goal has an honest path and the other has none, and they produce opposite teams.
Leaders reach for the second kind for understandable reasons. They feel behind, and a big number is the fastest thing to produce when you feel behind. It fits on a slide, and it sounds like a plan without requiring one. There is also research that seems to back it up. Edwin Locke and Gary Latham spent decades showing that hard, specific goals beat “do your best.” Plenty of leaders take that to mean harder is always better. The same research found the effect depends on people being committed to the goal and able to reach it. Leave those two conditions out and the finding means something different.
The goal with an honest path
A goal with an honest path is hard and has a route someone can name. “If we finish the queue migration in January and take two engineers off feature work for the quarter, we can cut pager volume in half.” The team might miss. The migration might take until March. But the work goes into the route, and if they miss, they know which part of the route broke.
That is the first benefit: a miss carries information. The second is who stays. Strong engineers like a hard problem with a real chance of solving it. That is the job they signed up for, and a goal like this keeps them in the room.
The cost lands on the leader. Someone has to do the planning work, and someone has to write down what stops so the route can happen. The two engineers taking on pager volume are not shipping features this quarter, and the leader has to say that out loud to the people who wanted the features. Sometimes the honest number is smaller than the one that would have looked good on the slide.
John Doerr’s account of OKRs in Measure What Matters gets one detail right that most companies copy wrong. At Google, as he describes it, hitting about seventy percent of a stretch objective counts as success. That rule makes a miss safe to report. A team that knows seventy percent is a good outcome has no reason to fake the last thirty.
The unrealistic goal with no honest path
A goal with no honest path runs the other way, and the cleanest case is a bank.
Wells Fargo pushed its branch employees to sell about eight products to every household. The target came from the top and ran down to people working the counter for an hourly wage. In 2016 regulators fined the bank $185 million after finding that employees had opened accounts customers never asked for. The bank said it had fired about 5,300 employees over several years for it. Later reviews raised the count of possibly unauthorized accounts into the millions.
It would be comfortable to blame 5,300 bad people, and it would miss the number that sat above all of them. A study published in 2004 by Maurice Schweitzer, Lisa Ordóñez, and Bambi Douma found that people working toward a specific goal they had not reached were more likely to overstate their performance than people told to do their best, and the effect was strongest when they were close to the goal. Decent people, a target with no honest route, and a short distance left to go add up to people filling the gap with something other than the work.
Engineering teams have their own versions. A test coverage mandate produces tests that run the code and assert nothing, and a velocity target gets you story points that inflate a little every sprint until a three is an eight. Tell a team to halve incidents and you will get a long debate about what counts as an incident. The newest version is “half our code will be written by AI agents by June,” set by a leader who has not given the agents access to the repo. Engineers paste a snippet from a chat window into a commit and tag it AI-assisted. The dashboard goes up, and nothing about how the team works has changed.
Charles Goodhart noticed this in economic policy in the 1970s, and the anthropologist Marilyn Strathern later gave it the phrasing everyone quotes: when a measure becomes a target, it ceases to be a good measure. The number keeps moving after it has stopped telling you anything.
Every target is supposed to point at something good. Pager volume points at engineers who sleep through the night and customers who can use the product. Coverage points at code that does what it claims. Gaming the number cuts it loose from the thing it was pointing at, which gives a leader a second test: if the number moved and the thing it points at did not, the team achieved nothing.
Often, before anyone games anything, the strongest people leave. Some stay and argue with the number, which is the best thing that can happen to it. They can do the math faster than anyone, they have options, and they do not want to spend a year on a treadmill that cannot move. The people who stay have three choices. They can miss and take the blame, burn out trying, or close the gap some other way. A leader who sets a target with no route has told them, without saying it, that the honest way is not expected.
The cost to the leader is trust, and it is spent before the quarter even starts. The team learns that the leader’s numbers are not real. Next quarter’s number gets discounted on arrival, including the honest one.
The real difference: a wish handed out as a plan
Aristotle draws the line that separates these two goals, in Book III of the Nicomachean Ethics.
He separates wish from choice. Wish, boulesis, can be for anything, including things that are impossible. You can wish you would never die. You can wish your team would win the championship, which is not up to you. Choice, prohairesis, is narrower. Aristotle says nobody chooses the impossible, and anyone who claimed to would be thought foolish. Choice is only about things in our power, and it comes after deliberation, which is working out the means to an end you already have.
Put that next to the planning slide. A target with an honest path is something a team can deliberate about: which work first, who moves, what stops. A target with no honest path cannot be deliberated about, because there are no means to work out. The leader has written down a wish and handed it to the team as if it were a choice.
Long odds do not turn a goal into a wish. A goal can be chosen if every step of the route is something the team can do or can be given, even when the chance of finishing is low. It becomes a wish when some step depends on something nobody controls, or on headcount and time nobody is going to provide.
The team still has to choose something, because they are still accountable. So they deliberate about the only thing left in their power, which is how the number gets reported.
None of this lets the person who cheats off the hook. The employee who opened an account in a customer’s name did something wrong, and so does the engineer who writes a test that asserts nothing. They chose it. The leader who built the treadmill made a different choice, and it is the one that made the cheat likely. Both are real, and the leader’s is the one that repeats across the whole team.
The person handed the number has a duty before it comes to that, at any level. On the first day, say in writing that you do not see a route and what a route would need. At the end of the quarter, report the honest number. A new hire and a staff engineer owe the same two sentences, and so does the director who got the target from a VP.
How to tell which goal you are setting
Most of this is fixable before the slide goes up.
Ask for the route before you announce the number. Find the two people closest to the work and ask them how they would hit it. Listen for an answer they would be comfortable giving you in writing. If nobody can describe one, or the only route needs headcount, time, or cuts you are not willing to give, you are holding a wish. Shrink the number or give them what the route needs.
Say what stops. A real stretch goal comes with a list of things the team will not do this quarter. If the list is empty, the plan is to do everything at once, and you already know how that goes.
Ask how a decent person would cheat. Before you commit, ask the team: if someone good were desperate to hit this by the deadline, how would they game it? They will know in about thirty seconds. Then pair the target with the number that would catch the cheat. Track coverage next to bugs that escape to production, and incident count next to outages customers report.
Make an honest miss safe and a perfect score a question. Thank the team that reports seventy percent with a clear account of what broke. When a hard number comes in at exactly a hundred, ask how, with curiosity. Most of the time there is a good answer, and the times there is not are the ones you need to know about.
Put the agents on the work. If your goal is about AI, the honest path runs through access. Give the agents the repo, the tickets, and the test suite. Pick three workflows where the model does the work end to end, and measure the hours that come back. A team with agents in the repo will usually beat a “half the code by June” target without anyone tagging a commit. A team with chat windows and a mandate will hit it on the dashboard and nowhere else.
I wrote last month about celebrating decimals, the habit of applauding partial progress as if it were the finish. This is the same problem seen from the other end. A goal the team cannot believe produces numbers nobody can trust, and then the measuring stops meaning anything.
Final Thoughts
A big goal with a route pulls a team forward even when they miss, while one with no route pushes the best people out and teaches the rest to make the number some other way. Both look ambitious on the slide, and only one is a plan.
Aristotle’s line is the test. You can wish for anything, but you can only choose what is in your power, and you can only plan your way to something that has a route. A leader controls two things here, the design of the target and what happens when someone reports a miss, and those two are enough. Before the next number goes up, ask the people closest to the work how they would hit it. If they cannot tell you, someone will hit it anyway, and you will not like how.
Setting goals a team can believe is a leadership skill, and it can be practiced. That is what MasteryLab is for.
