Cluster analysis

Shamu

Hi, 
I characterized data in respect to a taxonomy (business models: 5 dimensions a 5 characteristics): around 100 objects with mutually exclusive characteristics (each object has exactly 5 hits). So, I transponded the data already in binary code (1 and 0). I am already familiar to k-means clustering approach but not sure if this is the right approach in this application. My objective is to find certain archetypes (startups) in the data and name them. 
I already found a similar work, but they have used R for clustering and agglomorative bottom up hierarchical clustering with the multiscale bootstrap resampling approach from the PVClust Package. 
What would be your approach in this application? I prefer a two stage approach: I'd use Ward’s minimum variance method to determine the number of clusters, which I further need, in the second stage, for the k-means clustering.