Information-centric networking (ICN) is structured as a distributed caching system in which user requests arrive at edge nodes and traverse the cache hierarchy to access their demanded contents. If an intermediate cache node holds a copy of the desired content, it is promptly delivered to the client within the local domain. Otherwise, it must be retrieved from external sources, causing longer access latencies and higher provisioning costs. To minimize the external logistic costs from the backhaul, we harness the potential of Interaction-based Caching to preserve the most popular and largest contents for a maximum time in the local network. This strategy increases the cache hit rate and reduces the total volume of data that needs to be imported from external sources and the associated logistic costs. Simulative performance evaluations prove the significant gains of the proposed algorithm in terms of hit ratio (over 10.64 %), external volume import (over 1.18 times relative to internal volume transport), and minimum content provisioning costs (over 5.1 times relative to internal transport costs) with respect to the Leave Copy Everywhere (LCE) strategy.