Are the CryptoPunks dying? The data tells an interesting story.

As a small server owner and a data analyst, I recently grew tired of not having enough data to understand how my Discord server is performing. The de facto member count statistic is bland, so I started to brainstorm new methods to quantitatively measure how healthy a community really is.

We came up with a series of indicators to not only measure server activity in a holistic way, but also to use these insights to improve our server on multiple dimensions. The tool quickly became popular in my network, so we decided to share it by applying it on the CryptoPunks server, the OG project that began the NFT revolution! As of today, you can also use our tool to measure how your server is performing here.

TL;DR:

We’ve used detailed data analytics to take a deep dive look into the CryptoPunks Discord server. Our dashboard identified a couple red flags that looks to be eroding the underlying health of the community. If you’d like to give it a try for free, sign up here.

The Basics

As seen in the charts below, public interest in CryptoPunks (and NFTs in general) really increased dramatically in Q4 2021 and hit a local peak in late January this year based on daily messages exchanged.

Despite having more than 50k+ users on the server, CryptoPunks has a measly 100–150 unique users on any given day.

Most interestingly, the number of first time participants shows that while CryptoPunks is able to attract these initial users, number of messages failed to increase linearly with the user count. This suggests that on average, users are chatting less and less on the server. At peak, more than 50% of its daily users are first-time visitors, most of whom require extra attention from the server owner and moderators to build up retention.

Depth

The basic stats above spoke to the quantity of the community. It can be deceptive and mask any underlying weakness if you don’t look at the quality of user experience. Hence, we created more indicators to measure depth of experience:

This first chart shows the number of users spending substantial time on the server, calculated by looking at their message frequency throughout the day. The observation is that this stat is growing as more users joined.

Unfortunately, the real trouble is obvious once we look at this stat as a percentage. Despite the large influx of new users since October 2021, the percentage of meaningful users failed to increase and in fact, is lower than January 2021. As the server exploded overnight, CryptoPunk was unable to provide the same great server experience they were able to in the smaller format (a very common problem for a large server without deliberate prevention). As a result, a smaller share of people choose to participate in a meaningful way.

Red Flags

Distribution also matters. If there are 1,000 messages being exchanged daily on a server, we’d much prefer to see that it is created by many users rather than a select few power users. For the same 1,000 daily messages, we’d like to see that it is created evenly throughout the day rather than all happening in the 20 minutes.

To measure such distribution, we take inspiration from the GINI index, with a higher number indicating a more concentrated “wealth”.

In the below chart, we look at user uniformity. The higher the number, the more contribution power users have made. For any new server, the index is very low because there are no super users. Overtime, it becomes common-sight to see these power users develop and dictate the community pulse. We’d ideally like to see a number in the healthy range of [0.4 to 0.6]. Based on this one indicator, CryptoPunks is doing well.

Time uniformity is a simpler story. The lower the number, the better — it means your server is constantly active. Naturally, new servers have very high time uniformity indexes because your users are likely to exchange greetings such as “gm / gn” before going on with the rest of their days. For CryptoPunks, despite the massive influx of new people after October 2021, time uniformity index did not decrease significantly. This is a red flag. More users in theory should cause conversations to occur more evenly throughout the day, but this is not shown in the data.

Retention

We believe Discord communities follow the quintessential startup rule: grow or die. Based on our internal analysis, most servers have a retention rate around 0.6% to 2% after 30 days. Two of the largest factors determining this retention rate includes new user on-boarding as well as event planning, which requires a lot of community management. CryptoPunk has a retention rate of over 0.85%, which is close to the median server.

Using data analytics to understand communities is still within its infancy. Therefore, we plan to share our dashboard and make it available to all servers regardless of size in the future. If you are interested in giving it a try, sign up here: https://modraterhub.app/wait_list