D&D 5E D&D Beyond Releases 2023 Character Creation Data

D&D Beyond released the 2023 Unrolled with data on the most popular character choices for D&D. The full article includes a wide variety of statistics for the beta test of Maps, charity donations, mobile app usage, and more. However, I’m just going to recap the big numbers.

The most common species chosen by players are Human, Elf, Dragonborn, Tiefling, and Half-Elf. This contrasts with the stats from Baldur’s Gate 3 released back in August 2023 where Half-Elves were the most popular with the rest of the top five also shuffling around.

Also, keep an eye on the scale of these charts as they’re not exactly even. It starts with just over 700,000 for Humans and 500,000 for Elf, but the next line down is 200,000 with the other three species taking up space in that range. This means the difference separating the highest line on the graph and the second highest is 200,000, then 300,000 between the next two, 100,000 between the next, and finally 10,000 separating all the others.

Top classes start off with the Fighter then move onto the Rogue, Barbarian, Wizard, and Paladin. The scale on this chart is just as uneven as the last, but the numbers are much closer with what appears to be about 350,000 Fighters at the top to just over 100,000 Monks in next-to-last with under 80,000 Artificers. This contrasts far more from the Baldur’s Gate 3 first weekend data as the top five classes for the game were Paladin, Sorcerer, Warlock, Rogue, and Bard.

And the most important choices for new characters, the names. Bob is still the top choice for names with Link, Saraphina, and Lyra seeing the most growth and Bruno, Eddie, and Rando seeing the biggest declines from last year.

Putting that together, it means the most commonly created character on D&D Beyond is Bob the Human Fighter. A joke going as far back as I can remember in RPGs is, in fact, reality proven by hard statistics.

 

[Aldarc]

The same article says the most popular character name is "Bob." So by that logic people who hate stupid character names must be positively apoplectic.

Or maybe people don't really care what other tables do, and don't view disagreement with their own preferences as some sort of slap in the face?

In my experience, the use of "by that logic" almost always precede a logical ad absurdum. This post seems pretty consistent with that trend. @TwoSix has the right and good faith reading of the sentiment of my original post.

 

[Umbran]

No huge shockers there, especially once I noticed the sliding scales. Pretty good class parity, aside from artificers, which are not core to the PHB. Fighters used to be much farther ahead.

Check the scale on the vertical axis again - the scale changes as you go up.

 

[Oofta]

In my experience, the use of "by that logic" almost always precede a logical ad absurdum. This post seems pretty consistent with that trend. @TwoSix has the right and good faith reading of the sentiment of my original post.

Which is what ... that people that don't happen to share your preferences should be shamed? Made fun of? Even if someone doesn't like either one of those races, why would they care if someone else plays them? As far as the races I have no real opinion one way or the other. Meanwhile I don't care for butter brickle ice cream, but feel free to chow down all you want.

 

[J.Quondam]

The scale is a bit weird.

Yeah, i think "a bit weird" is an understatement. I can't figure it out. Is it logarithmic or something?

 

[Umbran]

Or maybe people don't really care what other tables do, and don't view disagreement with their own preferences as some sort of slap in the face?

I'm terribly sorry, but "people" is way too broad. Empirically, there's large numbers of "people" who do speak as if they do care about what other tables do, and what ends up in published books, and get very prickly when it doesn't match their preferred mode of game.

If you don't believe me, go start a discussion on "fudging"...

 

[Deset Gled]

Asking as a Roll20 player who has done this: Is it safe to assume a significant percentage of the "custom lineages" option is actually people using the customization options to re-create races from published D&D books because the user hasn't paid for the DLC to add it?

Gnomes you had your chance.

Time to replace them with goblin.

Also Half Race Appreciation Society recognizes Half Elf greatness.

If I had a say, I'd vote kobold over goblin. But I honestly don't expect either of those options would end up more popular than gnome.

 

[Oofta]

Yeah, i think "a bit weird" is an understatement. I can't figure it out. Is it logarithmic or something?

They change the scale when it gets below 100k, they kind of mashed two graphs with different scales together.

 

[Whizbang Dustyboots]

I'm mildly surprised that this table has stayed as consistent as it has. A lot of good new ancestry choices have been added in recent years.

Not a surprise to see custom lineages doing well, though, as they are a big gift for both roleplayers and min-maxers.

 

[Umbran]

Yeah, i think "a bit weird" is an understatement. I can't figure it out. Is it logarithmic or something?

It is piecewise linear. it goes by 20K per increment up to 100K, then by 100K per increment.

 

[Amrûnril]

Yeah, i think "a bit weird" is an understatement. I can't figure it out. Is it logarithmic or something?

As far as I can tell, it's completely arbitrary. And not arbitrary in a remotely useful way. The class figure expands the 0-100K range to fill 2/3 of the y-axis, even though there's only one data point falling into this range . The variation among the remaining 12 classes thus has to be compressed into the upper third of the graph.

As a result, the 12 PHB classes appear quite similar, with the Artificer trailing dramatically. Whereas a consistently applied scale, even a logarithmic one, would show that the Artificer is far closer to the Monk than the Monk is to the Fighter.

It is piecewise linear. it goes by 20K per increment up to 100K, then by 100K per increment.

Though the class figure breaks this pattern to include 90K, despite there being no actual data points between 80K and 100K.