Before step one: know which kind you are making
Two maps that look identical serve opposite purposes, and mixing them is the most common reason a map feels wrong without you being able to say why.
A capture map is for getting everything out of your head. Messy, fast, no judgment, ten minutes. Correctness is not the goal; emptiness is.
A structure map is for deciding how something is organised. Slow, edited, deliberately incomplete. Half the work is leaving things out.
Do not attempt both at once. Capture first, then structure — as two passes, ideally with a break between them. Trying to organise while you are still generating is what produces maps that are simultaneously overwhelming and missing the obvious.
1. Write the centre as a question
Not a noun. A question.
"Career" gives you nothing to push against, so the branches become categories — skills, jobs, money — and you have built a filing cabinet. "How do I move into work that doesn't need night shifts?" tells you immediately that some things belong on the map and some do not.
If you cannot phrase it as a question, you may not have a subject yet. That is worth knowing before you spend an hour.
2. Empty your head first, unstructured
Add everything you already think about the subject, as fast as you can, without arranging it. Twenty to forty items is normal. Do not create hierarchy yet — put them all at the first level even though that is obviously wrong.
The point is that you cannot see patterns in things still inside your head. Once they are all visible, the groupings tend to announce themselves, and they are usually not the groupings you would have picked in advance.
3. Group what emerged, then name the groups
Now look for clusters. Items that would fail together, or belong to the same concern, or depend on the same thing.
Name each cluster after you see it, not before. This ordering matters: categories chosen upfront force your items into shapes that were decided by convention rather than by the actual subject. Categories derived from what you actually wrote describe your situation.
Three to seven groups is the workable range. More than seven and you have not finished grouping. Fewer than three and one group is probably hiding something.
4. Cut
The step everyone skips, and the one that decides whether the map gets used.
Delete anything that is: already done, not actually yours, a repeat in different words, or something you know you will not act on. Being written down does not earn a place — a map that keeps everything becomes a monument to volume, and monuments do not get opened.
Expect to remove a quarter of what you captured. If you remove nothing, you were not capturing honestly in step two.
5. Deepen exactly one branch
Do not develop all branches evenly. Pick the one where more detail would actually change what you do next, and go one level down on that one only.
Even depth feels tidy and is usually wasted. Most branches do not need subdivision at all, and the ones that do reveal themselves as you work, not while you plan.
6. Draw the dependencies
Ask: which branch cannot move until another one has? Which one, if it went badly, makes several others pointless?
These are not parent-child relationships and should not be drawn as though they are. In Minty Path they are relationship lines — dashed cross-links kept visually distinct from the solid hierarchy, because "deployment depends on the code existing" is a different statement from "deployment is part of the code," and a map that conflates them is misleading you about its own structure.
This step regularly reorders the whole plan. Ten minutes here saves months.
7. Choose what is active
A finished map where everything is equally live is a list with better graphics.
Pick one to three branches as the current attention. In Minty Path that is Focus, encouraged as a deliberate guideline and kept independent of both status and dates — marking a branch Focus claims only that this is where attention goes, not that it is urgent or scheduled. Everything else stays visible and inactive, which is the honest state of most of any map most of the time.
When you are stuck at step one
Sometimes the subject will not resolve into a question and every category feels arbitrary. Three ways out:
Start from the irritation. Not "Health" but "why am I tired every afternoon?" Real friction produces real structure.
Start from the end. Imagine it went well. What is different? Those differences are your branches.
Start from the last time this failed. What went wrong then? Failure modes make excellent first-level branches, because each one is something that could fail independently — which is exactly the test a good branch has to pass.
Building it somewhere it stays useful
A map made and never reopened has done some good — the thinking happened. But the larger value is in the months afterwards, and that only arrives if the work attaches to the structure instead of migrating into a task app that knows nothing about the map.
You can build one now without an account and see whether the shape helps: the real Plan View and Work View, saved in your browser, and keepable if you sign up later.