Amongst the challenging problems that must be addressed in order to create increasingly automated electronic commerce systems are those which involve forming coalitions of agents to exploit a particular market opportunity. Furthermore economic systems are normally continuous dynamic systems generating many instances of the same or similar problems (the regular calls for tender, regular emergence of new markets etc.).The work described in this paper explores how simple forms of memory can be exploited by agents over time to guide decision making in iterative sequences of coalition formation problems enabling them to build up social knowledge in order to improve their own utility and the ability of the population to produce increasingly well suited coalitions for a simple call-for-tender economy.