Skill trees invite a specific mistake: treating the nodes as a shopping list of upgrades rather than as a set of commitments. A perk is not a bonus that sits on top of your character. It is a statement about which situations you expect to be in, and once a branch has been taken, the situations you are equipped for have shifted.
This is why two raiders with identical playtime and identical kits can have very different experiences. One has spent points on the branches that fit the fights they actually take; the other has spent points on the branches that sounded strongest in isolation. The tree does not have a correct answer. It has a correct relationship to your route.
The structure, before the advice
The official notes describe the tree as a progression structure with named skills rather than publishing a ranking. That distinction is worth holding onto: the game tells you what a perk does, and it does not tell you which one is best. Any page claiming an objective ranking is offering a playstyle argument in the grammar of fact.
Structurally, the branches sort into three broad kinds of payoff. There are branches that change what you can carry and how long you can stay out. There are branches that change how a fight resolves once it has started. And there are branches that change how easily a fight starts at all — movement, detection, noise. Those three categories answer different questions, and the mistake is to buy across all three at once.
Which branch first, and why the answer is boring
For a new raider, the branch that changes how a fight starts at all is worth more than the branch that changes how a fight resolves. The reason is arithmetic rather than philosophy: most raids are lost before a shot is fired, and a perk that keeps an engagement from beginning protects the whole raid, while a perk that improves the outcome of an engagement only helps in the engagements you were already in.
That does not make combat perks bad. It makes them second. Once a raider has a route and a habit of not being found, the combat perks start paying off reliably, because the fights that do happen are the ones that were chosen rather than stumbled into. Bought first, combat perks are insurance on a risk profile that has not been fixed yet.
The carry and endurance branch sits in between, and it earns its place once the route is long enough to need it. A raider whose raids routinely end because the pack is full or the healing is gone has a specific problem, and that branch is the specific answer. Buying it before the problem exists is buying a solution to an imagined raid.
Where the tradeoffs actually bite
Skill trees are interesting where they force a choice, and the choice here is between being able to leave and being able to win. The mobility and detection perks make leaving easier; the combat perks make winning easier. A raider cannot maximise both with the points available in the early game, and the correct allocation depends on which of the two the raid is currently failing at.
A useful diagnostic: look at the last five deaths. If most of them were engagements that started without warning, the failure is detection and the mobility branch is the answer. If most of them were engagements that were chosen and lost, the failure is resolution and the combat branch is the answer. If most of them were neither — the pack was full, the healer was empty, the route was too long — the endurance branch is the answer.
That diagnostic is worth more than any tier list because it is specific to the raider using it. It also updates: the deaths change as the habits change, so the allocation that was correct at raid ten is often not correct at raid fifty, and the tree allows a reallocation precisely because the answer is expected to move.
The trap of the near-complete branch
There is a specific psychological trap in skill trees: the pull to finish a branch because it is nearly finished. Three points in, the fourth looks cheap because it completes something. But the value of a perk is not its position in a branch; it is whether it addresses the current failure mode.
The way to resist is to hold points. A raider with two unspent points has options the moment the failure mode becomes clear. A raider with zero unspent points has to live with a decision made before the information arrived. Holding is uncomfortable, which is exactly why it works — the discomfort is the cost of keeping the decision available.
What the tree cannot fix
No allocation fixes a bad route. A raider who walks into the open with the best possible perks still walks into the open. This is the reason the tree should be the last thing adjusted when raids are going badly: the route, the kit and the objective are all upstream of it, and changing the perks to compensate for a route problem produces a raider who is well-equipped for a raid they should not be taking.
The reverse is also true, and worth saying: a good route with a mediocre allocation outperforms a bad route with a perfect one. The tree amplifies a plan. It does not replace one.
Reading a perk claim critically
When a guide says a perk is the best in the tree, the useful response is to ask what situation it assumes. Most perk arguments are really route arguments in disguise: “this is the best perk” usually means “this is the best perk for the route I run”. That is legitimate, and it is not transferable without the route.
The practical habit is to read perk advice for the situation it names rather than the verdict it delivers. If a recommendation does not name the situation, it is not giving you anything you can act on, because it has skipped the part that would let you check whether your raids resemble the ones it was written for.
The allocation that ages well
If one principle has held across the whole tracked patch record, it is that the perks which change what is possible age better than the perks which change a number. A perk that lets you reach a place you could not reach is still valuable after a balance pass. A perk that improves a damage figure is one patch note away from being a different value.
That is the argument for investing in movement and detection first, and it is also the argument for treating combat perks as tuning rather than as foundations. Foundations should be things a patch is unlikely to take away.