I actually was wondering about the general stats too, so I’ve been collecting data via the public API endpoints for the last month or so. My tentative plan is to keep doing this for 6ish months and get more accurate data, but I’ll use what I’ve collected so far and see if I can provide some ballpark stats. There are some caveats to the way I collected the data which may affect the accuracy, and I’ll address that at the end of the post.
updated average of people hitting level 60 per month
52 people on average reach level 60 a month. The median amount of time it took was 660 days, with a standard deviation of 482 days (the data is irregular though, so this isn’t very meaningful imo).
Average time between level 30-60
I haven’t had enough time to measure this, but based off of the average level-up time from below, I’d say about 350 days.
Number of users on level 59
There are 570 users who are level 59, 53 of which have reached level 59 in the past month. I suspect this number is inaccurate based on the fact that the forum levels will sometimes not match the level on the WK profile page.
Number of total reviews & lessons completed each day and the average correct / incorrect %.
The kanji / vocab that people struggle with the most (highest mistakes made).
The number of lessons per day could be estimated relatively easily, so I’ll keep this in mind for the future. The kanji or vocab item that people struggle with the most can also be found on the user profile (the wall of shame), but this would only be an estimation given that it we can only see their worst kanji+vocab item (there may be other items with the same % accuracy that aren’t shown). Everything else would be difficult or impossible to estimate without access to private user data.
Total user average time per level up
I calculated 16 days per level. Take this with a grain of salt though, because I haven’t been consistently monitoring level-ups.
What level do most users end up giving up? is there a wall that after they overcome they’re all good?
I looked at the levels of all users who haven’t levelled up in the last month, but I couldn’t really say anything conclusive. By looking at the percent decrease of users from consecutive levels, the highest percentages occur for levels 1, 4, 22, 37, and 43. However, the percent decreases were all over the place, so I can’t really say anything conclusive about this. That being said, this is one thing I’m very curious about, so I will be trying to model the likelihood that a user will stop at any given level, or continue on as a Markov chain.
Here are some more random statistics:
- There are 10,197 users that have levelled up in the last 40ish days
- There are about 14,700 level-ups a month
- There are around 29,000 lifetime users, and around 25,000 users with a paid subscription.
- There were around 750 resets in the past 40ish days (a good portion users have multiple back-to-back resets).
- The most common level to reset FROM was level 3. The most common level to reset TO is level 1. I’d like to look into these statistics more, specifically I’d like to investigate the likelihood that a user quits shortly after resetting.
As for the caveats, the data was collected from public Discourse APIs (the WK community). It seems like the level listed on profiles here doesn’t match their actual level in some cases, and may update at a later time, which skews a lot of the data I collected. Also, because I collected level data at irregular intervals, the time between level-ups isn’t accurate. I plan on addressing both of these very soon. If anyone is curious about any other statistics, I would be happy to look into it for you. Also sorry if there are any formatting issues as this is my first post on here.