Let’s get the first thing out of the way: I am not an authority, and I am not the Subclass Design Cops. If you have a design that makes you happy, go with God. This is just a collection of things I’ve either figured out or been taught by people smarter and more interesting than me.

The fundamentals: each class is very clear about the levels that its subclasses grant something, but it’s 100% okay to grant more than one something. Every class grants its first subclass feature by 3^rd level. In the 2014 rules, some classes start subclass features at 1^st or 2^nd level instead, but in the 2024 revision every class starts is subclasses at 3^rd level. This is controversial but, for whatever reason (such as me not working there), they didn’t ask me.

1. Whatever your subclass is delivering, it needs to deliver the core of its gameplay loop by 3^rd level. You’re going to build on that later, but the core experience needs to be up and running by 3^rd level. This is not negotiable – and the 2014 PH subclasses that fail this litmus test are failures. Looking at you, Archfey and Great Old One warlock.
2. It’s not a core gameplay loop if the player only does the thing once per adventuring day. You also don’t want to make their cool thing incredibly dominating but short-lived, as this rewards one-encounter-per-day adventuring (“the 15-minute workday”) too much. The Berserker’s Frenzy feature with its Exhaustion cost amounts to exactly this.
   • Once per short rest is okay, but think it over a bit.
3. It’s not great if the player doesn’t choose the use of their subclass’s main feature, though some players appreciate not having more to manage. (The Champion fighter’s expanded crit range that is their core feature.)
   • Features that kick in when you attack or cast a spell on a creature of a particular type aren’t going to play well, because the player can do absolutely nothing to influence whether the target is an Undead, Humanoid, Fiend, or whatever. In general, “Hunter of [Creature Type]” is not a great subclass pitch for this reason.
   • Features that hinge on a one-time decision and become passive thereafter are only pretending to contain a choice point. What they really contain is a push toward specialization. For instance, Weapon Specialization—a big thing in 1e through 3.5e, but not a thing in 5e, neither in feats nor subclasses. If your subclass concept is “great with a specific weapon,” what you’re probably doing is telling that player that the majority of weapon loot is uninteresting for them.
   • As a weird data point – in 14 levels of playing a Hexblade, I never rolled a 19 to crit against a creature that was my Hexblade’s Curse target. So one of the core pieces off the feature benefit didn’t come up even one time. I rolled 20s – never 19s. Dice are funny like that.
4. There’s probably no way to make a social interaction or exploration gameplay feature compelling enough to be the subclass’s main thing. The best features apply to combat and another area of gameplay. There’s nothing stopping you from giving them a separate feature, by 3^rd level or at a later level, that informs social or exploration gameplay. (Good social and exploration features are hard to write—a lot harder than most people care to admit, because a plain old bonus to a d20 roll feels shallow.)
5. Many classes have inherent multiple attribute dependency: paladins need Strength and Charisma, and all front-liners need Constitution. Monks need Dexterity and Wisdom, and so on. Be careful about subclasses that add another attribute dependency, or increase the need for an attribute that is probably going to be okay-at-best.
   • For example, when playing a ranger a few years ago, I saw that I was going to be stuck with Wisdom 12 unless I wanted to be even worse off in Strength (melee ranger, don’t judge me) and Con. I felt pretty well forced to go with Hunter rather than Monster Slayer as my subclass, as a result. And no, you can’t get around this in 5e by making Str, Dex, or Con a spellcasting stat.
6. If your subclass’s main change to gameplay isn’t a passive or at-will benefit, it has a use limit. That use limit might be an existing currency within the class (such as a monk’s ki pool, a spellcaster’s spell slots, a cleric’s Channel Divinity, or a bard’s Bardic Inspiration), it might be a plain 1/short rest (Hexblade’s Curse), a number of uses per long rest equal to your proficiency bonus (a lot of the subclasses in Tasha’s Cauldron of Everything do this).
   • If the base class leans heavily on getting everything back on a long rest (spellcasters other than warlocks, for instance), that’s even more reason to create subclass features that you regain on a short rest. The reverse is also true – fighters, monks, rogues, and warlocks all benefit from having per-long-rest features.
   • You can also create a new currency pool for the subclass, such as the Battlemaster’s Combat Superiority dice pool or the Rune Knight’s runes.
   • Especially for primary spellcasters, it’s fine and good to let them expend a spell slot to regain the use of a feature.
7. Pay close attention to the action economy of your core gameplay loop and all later features, and how that interacts with the action economy of the base class. To unpack that a bit, with examples: rogues and monks make heavy use of their bonus actions for base class functions. If your subclass grants a lot of things that also use bonus actions, the player can’t use their features well.
   • Likewise, don’t bother writing features for barbarians, fighters, paladins, or rangers that use their action – if you’re asking them to do something other than make their attacks with their action, something is probably going awry. As it is, spellcasting exists in some tension with attacking for Eldritch Knights, paladins, and rangers—and that’s why you see so many bonus action spells in the paladin and ranger lists.
   • This is your perennial reminder, apropos of nothing, that reactions require triggers. Don’t be the designer who writes a reaction and forgets to include a “When such-and-such happens, you can use your reaction to.” I have cornered the market on that error, and defend it jealously.
8. As you get into higher-level features, it’s okay if some of them are less vital overall, but it’s important that the player still feels excited to get them. At the same time, be careful about saving the very coolest thing for the last feature, because you’re saying the player either won’t get to that feature before the end of the campaign (most campaigns don’t reach 14^th level or above), or they won’t get much of the campaign’s length to play with that new toy.
   • If a feature at 14^th level or above needs to have a use limit, proficiency bonus per long rest isn’t a very interesting limit, because you’re starting at 5/long rest and stepping up to 6/long rest. PB/long rest works best if it’s boosting a sense of progression.
9. You can’t pre-empt the base class. For example, it would be cool to have a feature that built on a rogue’s Uncanny Dodge—“when you use your Uncanny Dodge features, this also happens.” You can’t assign that feature to 3^rd level, because the PC can’t use it yet. Because the rogue progression is weird, you can’t assign that feature until 9^th level. This is particularly an issue in classes that gain subclass features at 1^st or 2nd level – the class’s main gameplay features aren’t online yet, so you can’t modify them.
   • If the feature also does something else, you can write a feature that also does something else when a later feature comes in, but make sure you really really need to do that, because it is a bit awkward.
10. Be careful when giving a subclass something that is clearly a feature from another class. I’m not saying never—otherwise there would be no Eldritch Knight—but move carefully. (I still think Trickery clerics should have some amount of Sneak Attack progression, for instance.)
11. Design has trended away from this for reasons, but I do love subclasses that support a huge range of character concepts: everything from the Armorer artificer (tank armor and blasting/skirmishing armor) to the infinitely variable Battlemaster. I think the specific options of the Hunter ranger need some retuning, but the concept has a ton of promise.
    • Not you, Way of the Four Elements Total Landscaping monk. There’s no hope for you until the new Player’s Handbook. You were one of the only monk-specific things not sufficiently improved in UA releases.

Okay, that’s what I’ve got for now. Let me know if this is confusing or unclear. I hope it’s helpful, interesting, or at least provokes some thoughts on how to further refine some of the points I’m making.

Many thanks to Kainenchen for helping me fix the mess that WordPress made of this.

If you’ve enjoyed my writing, consider backing my Patreon! For as little as $1 a month, you can support my writing and get early access.

---

You can now buy Under the Seas of Vodari in PDF!

Under the Seas of Vodari is a 300-page book that provides 5E/Fifth Edition material for both gamemasters and players to dive into a world full of action and adventure set beneath the waves of Vodari or your own campaign world. This book is a follow-up to The Seas of Vodari, a setting book filled with swashbuckling action and magical adventure. Dive into your first adventure today!