Here, i am suggesting the use of market basket analysis as follows:
1) Getting user features (e.g. usage, demographic, etc.)
2) Segmenting customers into groups based on their features.
3) Identifying purchasing/spending behavior regarding previous offers.
4) Use association rule mining (or market basket analysis) to extract rules.
5) Use these rules to give predictions about the next best offer for each user.
Any suggestions/improvements/notes are welcome. Thank you.