I have watched the same scene play out at a dozen companies, including ours in the early days. Leadership spends a quarter writing an annual plan, presents it with great ceremony, and by March nobody can recite a single line of it. Meanwhile the team lives inside a daily task list that has plenty of motion and no direction. Both artifacts exist. Neither one is doing the job. The annual plan is too slow to react and the daily list is too short to mean anything. The space between them is where most planning quietly fails.
The fix is not a better template or a heavier process. It is three honest planning horizons that each answer a different question and feed the one below it. Where are we going. What does this quarter buy us. What does this week prove. When those three are connected, a person on the team can trace the task in front of them all the way up to the direction of the company, and a leader can look at the direction of the company and see it expressed in real work. That line of sight is the entire point of project management, and it is the thing almost every planning ritual forgets to deliver.
The long horizon: where we are going
The longest horizon is direction, and it should change slowly. I think of it as a twelve to eighteen month view, deliberately vague on dates and deliberately sharp on intent. Its job is not to schedule anything. Its job is to make trade-offs legible. When two good ideas compete for the same engineer, the long horizon is what tells you which one moves you toward where you said you were going and which one is just attractive.
Most companies overload this layer. They cram it with feature lists and quarterly milestones and call it a roadmap, then act surprised when reality redraws it within months. A roadmap full of dates is a forecast pretending to be a plan, and forecasts age badly. Keep the long horizon to a handful of statements about the destination and the bets you are making to get there. Three to five is usually enough. If you cannot say your direction in five sentences, you do not have a direction, you have a backlog with ambition.
This is also the right altitude for company strategy and, if you run them, the parent OKRs. Objectives belong at the long horizon because objectives are about where, not when. The mistake is stopping here, because an objective with no shorter horizon underneath it is just a wish that everyone nods at and nobody owns.
The medium horizon: what this quarter buys us
The quarter is where alignment is actually won or lost. The long horizon tells you the destination, but a destination does not tell anyone what to do on Monday. The quarter translates direction into a small set of outcomes you are willing to be measured on, and it is short enough that you can hold the whole thing in your head and long enough that real work fits inside it.
This is where OKRs do their best work, if you let them stay outcomes rather than collapsing into a task list with a new name. A key result is a number that moves, not a thing you ship. Shipping the new billing flow is a task. Cutting failed payments by forty percent is a key result. The difference matters because the first one is done the moment code merges, while the second one keeps you honest about whether the work mattered. We run a quarterly cycle with no more than three objectives per team and three key results each, and we treat anything beyond that as a confession that we have not chosen.
The discipline at this horizon is subtraction. Every quarter, more good ideas arrive than the team can carry. The plan that says yes to all of them is the plan that delivers none of them well. I would rather a team finish three things and feel the weight of having shipped them than start nine and end the quarter with a graveyard of eighty percent. Pick the few outcomes that buy you the most ground toward the long horizon, and consciously park the rest. Parked is not killed. Parked is a queue you revisit next cycle.
Example: suppose the long-horizon bet is that Atlas should become the place a team runs its whole operation, not just its tasks. The quarterly outcome that buys the most ground is not a flashy feature. It is getting median time-to-first-project under ten minutes, because onboarding is where the bet is dying. One number, directly traceable to the direction, and it reframes three months of work.
The short horizon: what this week proves
The shortest horizon is the week, and its only job is evidence. Each week should produce proof that the quarter is on track or proof that it is not, while there is still time to react. This is the horizon people are most comfortable with, because it looks like ordinary work, and it is also the one most disconnected from everything above it. A week of busy, unrelated tasks is the most common way a beautiful quarterly plan quietly dies.
So the test for the week is brutal and simple. For every meaningful piece of work, you should be able to name the key result it advances. If you cannot, one of two things is true. Either the work does not belong this week, or your quarter was missing something and you have just found it. Both are useful. What is not useful is work that floats free of any outcome, accumulating in the system and feeling like progress.
We keep the weekly horizon lightweight on purpose. A short list per person, each item tied to a quarterly outcome, reviewed once at the start of the week and once at the end. The end-of-week review is the part most teams skip, and it is the part that does the work. Did the week move the number it promised to move. If three weeks in a row produce no movement on a key result, that is not a motivation problem, that is a signal that the plan is wrong and the medium horizon needs to change.
Making the three feed each other
Three horizons are not three documents you write once. They are a loop. The long horizon sets direction, the quarter converts direction into measurable outcomes, the week produces evidence, and the evidence flows back up to tell you whether the quarter and even the direction still hold. Cut any of those connections and you get one of the failure modes everyone recognizes: strategy that never touches the ground, quarters that drift, or weeks that stay busy while the company goes nowhere.
The mechanism matters less than the linkage. You can run this with a spreadsheet and a recurring meeting if you have the discipline to keep the layers connected. What breaks people is connection rot, where the quarterly plan lives in one place, the weekly work in another, and the strategy in a deck nobody opens. The reason we built planning the way we did in Atlas is that a task should carry its lineage with it, so anyone can see the outcome it serves and any leader can see the outcome expressed as real, moving work. When the horizons share one surface instead of three, alignment stops being a meeting you hold and becomes a property of the system.
If you take one thing from this, make it the test, not the framework. Look at the work on your team's plate this week and ask what each piece proves about the quarter, and ask what the quarter buys toward where you said you were going. If you can answer cleanly, your planning is working, whatever tools you use. If you cannot, no annual offsite will save you, and no clever daily list will either. The horizons only work when they feed each other.