Need direction for a retail customer cluster analysis

I am intensely interested in learning about machine learning techniques and I thought this would make a good project to learn on.

I have access to customer transactional data down to the SKU level with product hierachies. I have access to customer geographic data. I do not have demographic although I suppose I could append zip code averages based on census or other public data if this sort of thing is appropriate.

The end goal is build a group of customer profiles based on what's available.

I need help in deciding what kind of clusters/segments to attempt and what insights they could potentially bring. Should I use K-means/ neural net/ etc...

Tools at my disposal: SQL, SAS, R (beginner).

BONUS QUESTION: What could I do with demographic data that I can't do with the resources above? Demo data is expensive to purchase but perhaps I could justify the expense to my employer.

Any direction would be much appreciated.