After our last post about fashion brands in web3, we were asked to do an analysis of RTFKT, Nike’s web3 arm. And being the great listeners that we are, we did just that by looking into their flagship collection, Clone X. This week, we’re highlighting Ludis’ Clustering Algorithm App and how it can be leveraged by enterprises to better understand customers and create strategies to engage with them.
How enterprises can leverage this type of analysis:
RTFKT can use the findings from each insight to target customers matching the Clone X whale customers profile or find ways to make existing customer clusters behave more like whales.
Clustering Parameters & Focus:
We will be focusing our insights on the highest value cluster of the 5 identified after running our app along wallet value and transaction count variables. This cluster, which we call Whale Customers, is characterized by having the highest wallet values (Combined value of ERC20, ERC721, ERC1155 contracts) and the highest transaction activity.
Focusing on the highest value customer group, the Whale Customer Cluster, we can see:
We found that as you go from the lower value customer clusters to higher value clusters, there’s proportionally less value allocated to NFTs. The whales also own the largest number of NFTs compared to any other cluster. This is interesting as we’d love to dig deeper and see what’s causing this trend. Do whales own assets that maintained their value during the crypto sell off? Did they sell off assets earlier to prevent value loss while other customer groups held? Our guess is that whales have a greater ownership of “blue chip” NFTs that kept their value, because of the next point that we’ll discuss.
We take a look at the second finding, average NFT value within a wallet. Our whales owned an average of 1,250 NFTs equating to $56 per NFT in their wallet. This might be a result of them owning more NFTs that maintained their value compared to other assets. If this weren’t the case we would’ve expected average NFT values to be in line with the other clusters. However, the lower value clusters had average NFT values about 35% lower in the $30-$40 range. So there must be some high value NFTs in the wallets of whale customers that brought up the average value, especially when whales have 7x more NFTs than any other cluster. We quickly looked at each cluster to see if they happened to own any Bored Apes Yacht Club NFTs, which has a floor price of 68.9474 ETH (~$88K) on OpenSea and has remained relatively steady in price point when compared to many assets that went under. It turns out the whales do have a higher proportion of Bored Ape Yacht Club owners at 12%.
So are the whales holding onto their valuable NFTs and trading other assets? Well our third finding seems to suggest that’s the case. Despite having 9x more transactions than other customer clusters, the proportion of those transactions that are NFTs is the lowest at 31% vs 59% as the max within another customer cluster. They seem to be focused on trading their ERC20 crypto coins while holding onto NFTs. It will be interesting to look at the average hold period for NFTs between the clusters as a future analysis to further explore this point.
These findings tell us that RTFKT should find customers outside their collections that look like Clone X whales as they’ll have high levels of liquid wealth, high engagement and greater retention.
So how can they do this? The first step is to sign up for the Ludis Platform and we’ll help with the rest. Reach out to us at www.ludisanalytics.com to get early access to this Clustering App and many others.
Ludis Analytics’s proprietary data analytics for blockchain tools. Data as of December 6th, 2022. For more information please visit our webpage and contact at www.ludisanalytics.com
Ludis' Web3 Clustering Methodology:
Ludis’ proprietary clustering algorithm takes all wallets owning the specified collection token and clusters the wallets along either 2 or 3 dimensions. The application is currently able to run on ERC20 contracts as well as NFTs (ERC 721 and ERC 1155). For this analysis we used the following 2 dimensions:
First, we segment a collection’s customers by identifying all the wallet addresses that own the asset and collect the respective value and transaction activity. We analyze the transactional behavior and holdings and run our algorithms against this data set to determine the customer groups.
Clustering algorithms take disconnected data points with each as its own cluster. Then they connect the two closest points creating a new cluster, repeating this process over and over to create larger clusters. For this analysis we chose 5 clusters as it has the lowest sum of squares within each cluster.