On Skill (or Talent) Trees and How to Make Them

In some video games the in-game capabilities of a character or controllable/useable unit are catalogued as ‘attributes’, ‘skills’, ‘perks’ and others. They are meant to give the player the ability to choose what the aforementioned unit can or cannot do. Such progression systems are often found in RPGs, though they have branched out into other genres as well. Skill Trees are the acorn that didn’t fall far from Tech Trees. No more puns, promise.

The modern tendency is to implement RPG-specific mechanics into other mixes, such as FPS or Tower Defense, without the substance of story-related choices, exploration or immersion. One of the most used said mechanics is the Skill Tree – Watch Dogs has it, Shadow Warrior has it, Dead Island has it. Good old purists would say that it has become a plague to the core shootey experience of older titles.

Despite being used so very often, you rarely see a tree structured in an impressive and unique way. Most offer a few nodes with scaling passives and some abilities, every now and then, ordered in tiers by level. We do get some breaks, some fresh air, some bubblegum scent due to titles such as Path of Exile, Diablo III, Bastion even.

As RPGs are by far my favorite genre and the main one I want to dive into designing and developing, I read as much as I could, analyzed as much as my mind allowed and designed and designed as much as my poor soul afforded. Here are some general ideas I decided to share, which, I do hope, will help you in your work.



After all, it’s the first thing you see – and oftentimes it is the same one: a few branches. Not only that, but they generally come in triples. Sometimes they are disguised, but we all know, dear Witcher 2, those are talent trees at their almost-generic finest.

If you feel like trying something different, take a good look at what Path of Exile has to offer. All characters have one big tree, but each class starts at a different point. Marauders are closer to nodes related to Strength and Melee Damage, but it takes a low amount of levels to reach the other sides of the tree.

Dragon Age II and Dragon Age Inquisition offer branches with a more interesting layout. The main nodes can have more than one ‘in’ and one ‘out’ connection and they can have smaller nodes which offer bonuses to the main ones. This ‘atomic’ structure allow the player not to invest into a skill or an ability more than they do desire (in order to use it, or in order to further progress through the tree).

Another way you can go around it is not having connections between the nodes at all: Diablo III is a good example here.

If you do want your talent tree to have a unique shape your main concern is not to make it look like this:






Before the Cataclysm expansion, players were free to add points into whatever specialization they chose. Cataclysm made it so that a single specialization was to be chosen first, limiting the investment possibilities to a single branch until you reach a certain amount of points spent. Mists of Pandaria reshaped the tree completely: there are tiers corresponding to certain levels and each of them has 3 skills. The tree varies based on the class and specialization chosen.

Blizzard, in my opinion, tried to eliminate dump skills and make choices feel significant. One way they did it was, at first, having to choose a specialization. Choosing one gives new skills the moment it is chosen and denies access to other specs until a certain level is reached – this means that your choice is important, both for now, and for later. Mists of Pandaria let players forget about most restrictions.

One basic restriction almost all talent trees in video games have is that if you choose the talent A now, you have to wait until you get to choose the talent B. Mists of Pandaria made it so you could always change between the talents A, B and C, those being talents of the same tier. This also removes another restriction: having to unlearn a talent if you are not satisfied by the build.

Talent-choice related restrictions also come from the links between the nodes. You can’t possibly cast a Blizzard spell unless you got unlocked the Frostbolt spell, now, can you? A piece of advice I give you on this end is to always try to use an ‘or’ gate. Being able to cast Frostbolt or Frost Nova is enough to learn Blizzard, it doesn’t matter which one you choose. Sometimes, you might find it interesting to use an ‘and’ gate as well: in order to learn Fireball you’d have to learn Arcane Missile and Fire Mastery.

A certain type of restriction I don’t recall ever seeing is using other types of logic gates. Things like: you can’t learn Blizzard if you have Meteor Shower unlocked, or you can learn Healing Light as long as you have no dark spells learned, or you can learn Berserk only after you learned other 3 Warrior skills.

Other restrictions you often find are: character levels, the number of points spent in the tree, stats (such as Intellect, Stamina) and, most obviously, points. All of these are elements which modify the pace and feel of the game. Being unable to unlearn or having a skill restrict another means you want your game to have a great amount of replayability. Being able to invest more points into a skill to increase its potency means you want the player’s progress in terms of gameplay to stall, without making the power of the character to be left behind.

Try to analyze every single restriction you find in the talent trees of the games you played and see how they change the pace. There’s no better way of understanding it than by starting with your very game experience.



The main restriction the player faces when going up (or down) the tree are talent points. Skill points or talent points are the most used currency for unlocking or upgrading nodes. They often come when leveling up.

Sometimes it can get more creative than that: Bastion uses Fragments (money, basically) in order to upgrade weapons with special passives, alongside a certain ingredient unique for each weapon.

Grim Dawn and Titan’s Quest had it another way: you get 3 points when leveling up, which you get to spend into skills. However, in order to unlock the next tier, you have to spend the very same points to fill a bar with general stat lines (such as Health, Magic Power etc.)

Skyrim made it that the perk allocation is no different from most games, but leveling up means actually using skills rather than gaining experience through quests or slaying monsters.

When it comes to resources there are two main questions you have to ask yourself: how much can you spend into one individual node and how easily and often can you spend?

Talents with more than 3-5 maximum points mean that you want the character to objectively progress through increasing stats (making a Fireball deal more damage at Rank 5 than Rank 4). Talents with fewer maximum points (even 1) mean that you want them to add a new ability or trait which influences the gameplay (being healed whenever you deal a critical strike, for example).

Based on the way you shaped the tree and the resource capacity of the nodes in it, you have to decide how often you can spend. Is getting a skill point whenever you level up enough? How much of the tree will you have complete at max level? If you use a resource apart from skill points, how easily can they be acquired? Does the way your leveling system or resource system make certain builds easier to play? Does the way you made the tree make it easier for the player to gain points at the beginning of the game?

As with all things, experimentation is the one thing that will teach you what is suited for your game. Do not be afraid of using things such as Gold or even in-game time as resources required for advancing through the tree.



One thing RPGs do is making all nodes you can ever unlock visible, with few exceptions. This encourages the player to value tree planning over exploration. One idea I explored a little bit is making most nodes in a talent tree invisible. A node has possibly 4 states: locked, unlockable, upgradeable and maxed. What if all nodes that are locked are invisible until they become unlockable?

Imagine you are playing a rogue-like and you have a big empty talent tree except for three nodes far off from each other. You can unlock Cut, Fireball and Heal. As you unlock Heal, two more nodes linked to Heal are revealed to you: Bless and Turn Undead. As you drink a potion, further on the game, you unlock the Rain spell, which is somewhere on your talent tree. Unlocking it reveals another node between the Rain spell and the Heal spell: Healing Rain.

This was the sort of idea I wanted to experiment with, in my case, in the context of a pure magic-theme. Rogue Galaxy did play with visibility: unlockable nodes require certain reagents to be unlocked. Once unlocked, they also reveal what they do. Locked nodes didn’t offer any kind of information about the node.

What you do have to keep in mind is what you want your talent tree to offer. Do you want it to offer exploration and discovery, or do you want it to make it all clear straight from the beginning the ways you can invest in your build?  The decision regarding visibility affects the replayability of the game, the player’s capability to plan ahead, the risk of making players feel cheated when they pick something they don’t like and forcing unlearning nodes to be harder (since exploration of the tree would be too easy otherwise – what’d be the point of spending all points then reusing them to explore another spec and so on and so forth? It would fake exploration, since people would get to explore the tree in order to plan how to build).


Now that I covered some ground about the nodes, it is far from enough. You came here because you either want to design a skill tree or because you were curious about what I wrote. In both scenarios, I have a few tips on how to design what the nodes contain. We’ll start with the numbers.

Imagine the following scenario: you have a spell called Ignite which causes AoE fire damage. You also got a talent in your talent tree which reads: Ignite causes enemies to burn, dealing 30 damage over 5 seconds. Sounds decent, doesn’t it? You explode them and then they burn.

There are a few possible scenarios. The first one: all your enemies are at full health and you cast this spell on them. They do receive more damage, but the talent doesn’t feel like anything else apart from a scaling passive. All it did was deal more damage than before, after 5 seconds. The second scenario: the explosion doesn’t kill anyone, however half of your enemies die because of the damage over time. This feels a lot more satisfying because it shows quite visibly what that talent is about.

In the case of this talent we learn that using the spell as an ‘execution nuke’ is a lot more satisfying than using it on enemies with high health. What if we changed the talent? Ignite causes enemies to burn, dealing 12% of their missing health as fire damage over 5 seconds. This means that the damage over time will be less powerful than the previous flat damage on high health targets, however it gives the spell Ignite a new purpose: it becomes an execution tool.

If you want to play around with numbers in your skill-designing adventures, I advise you to know that all numbers are powerful.

  • Flat numbers (20, 30, 157) offer constant effects, thus easier to predict effects – which is not something bad.
  • Scaling numbers (your Dexterity, your Level, your Movement Speed) are a way to make spells be useful later in the game without having them to be upgraded in any way. They also allow players to focus on a certain stat which increases the power of their skills (Mages like the Intellect stat, Warriors the Strength stat).
  • Dynamic numbers (your enemy’s missing Health, your remaining Mana) are the ones that offer the greatest variety of effects. They also make certain skills to be very specific in their role, just like how the talent I made up earlier encouraged Ignite’s role as an execution spell rather than a general damage-dealing tool.



Perhaps one of the most important things regarding the skills you design. The same good old Fireball spell gets overly overused – it does. You should not be afraid of exploring some quirky ideas. Even more, you should, by all means necessary, explore quirky ideas.

Let us presume that you have an Arcane Blast spell and you want a talent which offers survivability. ‘Whenever you hit an enemy with Arcane Blast, you are healed 20 health’. If your character’s health is increasing in time, then the effect will get useless. ‘Whenever you hit an enemy with Arcane Blast, you are healed 8% of your missing health’. Now that is something entirely different – it will cause scenarios in which Arcane Blast will save your hide, whilst limiting the heal you receive when you have higher health (when, you could argue, you don’t need the heal that much).

We can change it further. What talents do is that they change the playstyle. All talents change playstyle. Sometimes including a minigame or a special mechanic makes it even more interesting – it’s part of playing with risk and reward. ‘Whenever you hit an enemy with Arcane Blast, an orb is spawned nearby. When picked up, you are healed 12% of your missing health’. It is not a buff nor a nerf. It changes the way you play: it encourages kiting, since you don’t want the orb to spawn somewhere where there are enemies, it encourages playing in open spaces, so that the orbs are in reach. There will be scenarios when that 12% makes a difference, where the 8% couldn’t, and there will be scenarios when the minigame will prove unreliable for the player.

You, as a designer, have to know when it is beneficial to add a minigame and when it is not. As for the minigames you can add, there are endless possibilities: resource management, moving to some spot, avoiding some spot, playing around with cooldowns, self-debuffs etc.



Playstyle is a topic highly linked with the previous one. To be frank, all the topics from this article are linked with one another. From my experience, there are two ways you can go about making a skill tree: you have a skill idea and build a playstyle around it, or you have a playstyle idea and create skills for it. You can always change between them as you develop your skill tree.

In the case of a design I worked on, I had a playstyle in mind: a mechanic called Heat. Casting spells generate Heat, which vents over time. If your Heat reaches a certain limit you Overheat, exploding and losing all Heat. Around it I created passives which benefit from high Heat, thus allowing players to risk being close to Overheat for higher damage. Another passive was having the Overheat explosion also damage nearby enemies, thus offering another benefit for that playstyle.

In another case, I had the idea of an ability: you shoot a bullet and leave behind a shadow which does the very same ability after 3 seconds, then it disappears. I loved this ability so much that I build a talent tree around the mechanic of leaving shadows behind which mimic your initial ability. I found that the skill tree I sketched had two main playstyles: spamming shadows and acrobatic gunplay, which both feat the thematic of the Pistoleer/Demon Hunter I imagined.

When you want to start with the playstyle fantasy, be sure of two things: you have played a lot of games, especially ones in which what you played is similar to the playstyle desired; make sure that the talent tree you build offers more than just one playstyle. It would be boring not to have it offer more than a single playstyle. In Dragon Age II you could use Arcane magic in more than just one way, in WoW Druids on Feral got two forms with different playstyles – despite being the same branch, the same tree.

When you start with the abilities, skills, passives, make sure that they synergize. You rarely would want a passive that makes your spell deal 20% more damage when the targets are near you and one that makes your spells burn enemies when they are far, at the same time. A player doesn’t (always) get to max out all the nodes, which means that he will make choices. The choices he makes should feel empowering, should feel worth taking and able to work together with previous and future skill-choices.

One thing I advise is to think of several playstyles when making a tree and see which playstyles can stay, which can be mixed together, which can be thrown away. The same goes for abilities and passives – make several of them then choose what stays. I used to think of a general playstyle then create the nodes from the lowest tier to the highest, thus trying to understand what kind of development the player would want. What I do now is create several nodes, write them down, pick some which I feel are suited for the playstyles the tree intends and also feel satisfying to pick, then decide which go to what tier.


Another Kind of Restrictions

This last part isn’t about the talent tree itself, nor about the resources or mechanics of skills, but about the amount of skills you can have. One idea I’ve been experimenting with and really want to see take shape is the following. There are 4 types of skills:

  • Spells: You press a button and they do something, like a Fireball, or Whirlwind. You can unlock as many as you want, but can have only 4 equipped at a time. You can always change which ones are equipped.
  • Passives: They always are active once you spend points into them. They aren’t game-changers, but add to the general playstyle of the tree.
  • Traits: Traits are powerful passives which alter your playstyle significantly. Things like ‘you deal more damage when on low health’, or ‘casting a spell gives you a stackable Magic Power buff for 3 seconds’, or ‘casting a spell create a sand soldier which charges at a nearby enemy’. You can unlock as many as you want, but you can have only 3 active at a time.
  • Attack Modifiers: Passives which alter your basic attack. Things like ‘you throw fireballs instead of hitting people with a staff’, or ‘you gain a short burst of speed whenever you attack’. You can have only one active at a time, but since basic attacks are the most used attack, you’re going to use it a lot. Just the same as with Traits and Spells, you can have more unlocked, but only one active at a time.

Such limits prevent spamming abilities when higher level (unless you make the spells spammable as part of the gameplay intended) and having a fuzzy, unclear playstyle when maxing everything (the same way I did whenever I played Skyrim and got to be good at everything).



Are talent trees something suited for your game? That is something left for you to decide, discover and toy with. If you do want to create a talent tree, then please do cure us of the generic talent trees we see so very unnaturally often.

If you know that you’re going to design RPGs or things with abilities or talent trees in the future, I also advise you to take some time daily to design trees and skills. I myself filled countless pages and a few notebooks with diagrams, words, mechanics and what not.

I have to beg your pardon for the packed information, the casual remarks and everything else that might have disrupted your flow of reading. Perhaps I have to invest some more skillpoints into that.

You can find some more things on skills here.

6 thoughts on “On Skill (or Talent) Trees and How to Make Them

  1. Excellent article. I’d like to really emphasize a specific point in your article: the game should ALWAYS be clear on how it intends you to play.

    As Stormmer says, if your design is intent on the player exploring and discovering options – make that clear by suggesting (early on) that the player’s choices could be changed later in the game. Include what the costs/tradeoff for doing so is – in terms the player can understand (they may not know what 20 Valor Tokens are at the time they’re allocating their first few points).

    If your design is more about long-term planning and permanency – make that completely clear! Tell them the choices are permanent in a noticeable way, and give them the tools they need to do long-term planning.

    Liked by 1 person

    • Thank you a lot for your input!

      I think that a lot of games face the problem of not exactly knowing what they are or what they should be. Too many designers seem to want to implement as many shiny toys as they can, to make a title more attractive, to make it appear ‘modern’. It’s unfortunate that not all games are true to their essence.


  2. Pingback: On Designing Skills and Abilities |

  3. Pingback: Retrospective Journey – Ability Design | Stormmer's Gaming

  4. Pingback: Keys to Meaningful Skill Trees – GD Keys

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s