Ask three people on a product team whether a piece of work is done and you will often get three answers. The developer says yes, because the change is written. The tester says almost, because one case is open. The leader says they do not know, because they cannot yet see what is different for a customer.
None of them is lying. They are using different definitions of the same word. When you set direction, the definition that matters most is the one that only you can supply.
Output is easy to count
Tickets closed, features shipped, pages built, releases cut. These numbers are easy to collect and easy to put on a slide. They also feel like progress, which is exactly the danger.
Output tells you how much was produced. It does not tell you whether anything improved. A team can close a hundred tickets in a quarter and leave the customer’s experience exactly where it was. A team can also close ten and change how people use the product. If your only measure is output, you cannot tell those two quarters apart.
This is not a criticism of counting. Counting is useful for planning. It is a poor definition of done.
Done is a change someone can notice
Here is a definition that holds up under pressure: work is done when a specific person can notice a specific difference.
The person might be a customer who finds the checkout quicker to finish. It might be a support colleague who stops receiving the same complaint. It might be you, opening the product on a Monday and seeing something you asked for. The difference might be small. It still has to be something you could point at.
If you cannot name the person and the difference, the work is not yet described well enough to be done. That is a useful thing to learn early, before anyone has spent a month on it.
Three questions that turn a task into an outcome
Most vague work can be sharpened with three questions, asked before it starts and again before you accept it.
Who is better off, and how will we know? Name the person. Then name the sign. A shorter wait, a question that no longer gets asked, a task that no longer needs a workaround. Signs that live only in a status report do not count.
What would make you say no? Decide in advance what a disappointing result looks like. If you only define success, every result can be argued into one. A clear refusal condition protects you from the slow slide into “good enough”.
What is yours to decide, and what is not? Some choices carry a judgement about the direction of the product. Keep those. Others are craft, and can be trusted to whoever is doing the work. Saying which is which at the start saves both sides from the awkward discovery halfway through.
None of these questions needs a template or a meeting. They need a few sentences, written down where others can read them.
Why leaders skip this
It feels like overhead. The work is obvious, the team is capable, and a long description seems to insult both. So the definition stays in your head, and the team fills the gap with their own reasonable guess.
Sometimes the guess is right. When it is wrong, the cost shows up late, in a review where you say “this is not quite what I meant” and everyone is quietly frustrated. You are frustrated because the result missed. They are frustrated because they were never told where the target was.
There is another reason, and it is less comfortable. Defining done as an outcome exposes you. If you write down what should be different for the customer, you can be wrong in public. A vague brief can never fail. A clear one can. Leaders who accept that risk get better results, because clarity gives everyone something to correct.
What changes when done means an outcome
The first change is in the conversations. Status meetings stop being a tour of activity and become a short check on whether the difference is arriving. You ask “is it noticeable yet?” instead of “how far along is it?”
The second is in the handoffs. Work that carries a clear definition of done can be passed to someone else, another team member or an outside partner, without a long explanation. They know what to aim for, and you know what to look for when it comes back. A handoff without that is a hope.
The third is in your own attention. When done means an outcome, you review results, not activity. You do not need to follow every step, because you are not trying to infer success from effort. You look at the thing, and you decide whether it did what you asked.
The fourth is easy to miss: it becomes easier to say no. A result that misses the stated difference can be declined without it being personal. It is simply not what was agreed. Teams usually prefer that to the alternative, where approval depends on mood.
Where outcomes can mislead
A word of caution. Outcomes are not a licence to ignore everything else. Some work has no visible difference on the day it is finished and still matters: repairs that prevent a problem, upkeep that keeps a product healthy. For that work, the outcome is an absence, and the sign is a quiet year. Define it anyway, even if the sign is “nothing happened”.
Outcomes can also be set too far away. “Grow the business” is an outcome and a poor definition of done. Done has to be close enough that a piece of work can plausibly cause it, and a person can see it within a reasonable time. If you cannot connect the work to the sign, choose a nearer sign.
A small place to begin
Pick one piece of work that is in progress this week. Write two sentences: who should be better off, and what they will notice. Add one sentence on what would make you refuse the result. Share it with whoever is carrying the work, and ask whether it matches what they thought they were building.
Expect some surprise. The gap between your sentences and their understanding is the gap that would have cost you a review later.
Setting direction was never only about choosing where to go. It includes saying how you will recognise arrival. A leader who can do that is easy to work for, easy to hand work to, and hard to disappoint by accident.