90% of D&D Games Stop By Level 10; Wizards More Popular At Higher Levels

D&D Beyond has released some more data mined from usage of its platform. A couple of weeks ago, it published some stats on the most viewed D&D adventures, from Dragon Heist and Strahd all the way down to Rise of Tiamat. This time, it's a look at player characters by tier of play.

​

Tier 1 is levels 1-4, Tier 2 is levels 5-10, Tier 3 is levels 11-16, and Tier 4 is levels 17-20.

Tier 1 contains the most characters created on the platform (as you would expect), followed in order by Tiers 2-4. About 90% of games do not make it past the 10th level mark, as the developer notes.

​

This chart shows that the fighter is the most common class at all tiers, followed by the rogue. At third place it switches up a bit - the wizard becomes more popular in Tiers 3-4 than in Tiers 1-2, while the cleric and ranger both have a strong presence at lower levels but drop off at higher levels.

You can find the report in the latest DDB development video below.

[video=youtube;4tuIrGLKSik]https://www.youtube.com/watch?v=4tuIrGLKSik[/video]​

 

[dave2008]

yes, but none before 5e explicitly addressed the issue in terms of game design. bounded accuracy should address this in 5e, and according to the data it seems not to be able to.

Actually 4e addressed it much better than 5e, though it cheated a bit. And like all editions you had to really buff epic level solo threats to make them work

Never mind: I just noticed I was responding to a most that was 6 years old!

 

[Staffan]

Can we presume that they are not (and reasonably, can't) removing the vast amounts of low level characters created for campaigns outside of D&D Beyond or for expirmentation/fun from the data set? Of course most characters made on D&D Beyond are going to be low levels.

DDB folks have mentioned that they have methods that control for that – basically checking if the characters are actually used online or not. Things like seeing if checkboxes get marked for being used, hit points going up and down, and so on.

It likely won't catch characters that are generated using DDB and then printed out and used offline, though.

 

[Scribe]

What explains that data other than "people are playing a thing despite being unsatisfied with it"?

Yeah, because people want what they want.

Non-Magical Melee will always be popular. Now, the next step is people also unfortunately always want more.

 

[UngainlyTitan]

DDB folks have mentioned that they have methods that control for that – basically checking if the characters are actually used online or not. Things like seeing if checkboxes get marked for being used, hit points going up and down, and so on.

It likely won't catch characters that are generated using DDB and then printed out and used offline, though.

It should be quite trivial for them to tell. At least enough to be of use to them in market analysis.
First off, some folks make a big deal that D&DBeyond is self-selecting but when the sample size is in the tens of thousands, or higher then the sample size is big enough to be representative and it is at least representative of the market Wizards is most interested in. The people that will use stuff on D&DBeyond.
As for telling the difference between people making random characters as distinct to played characters. The played characters will be level up over time. Both the update interval and level interval will inform them of the reasons for the update. As will inventory management interactions between levelling events.
That is, a character getting periodic levels at one level at a time and having inventory changes can pretty reliably counted as a played character. Even if the character is played off the D&DBeyond platform (That is at the table or one another VTT).

I have been checking on D&DBeyond and I can conform that every dice roll made in a campaign I am involved in is still on their system as least as far back as last November, it really slows down when scrolling through the chatlog and my guess is that the entire campaign is stored. I will attempt to check later and edit this comment if I remember.

Edit: I was able to view the log back to September last year in a campaign that was about a year older than that. I do not know if this limit is due to what is stored in D&DBeyond or what the relevant buffer on the browser for the log information.

 

[schm0]

DDB folks have mentioned that they have methods that control for that – basically checking if the characters are actually used online or not. Things like seeing if checkboxes get marked for being used, hit points going up and down, and so on.

It likely won't catch characters that are generated using DDB and then printed out and used offline, though.

I ask because in past data analyses they had basically gone down the list of most popular subclasses and they were all the ones from the SRD. In other words, folks chose them because they didn't have anything else to choose.

 

[Demetrios1453]

Um, everyone, if you look, this thread was necro'ed by accident, and we already have several active threads on this matter. It might not be a bad idea to shut this one down, as it was from six years ago...