When sizing a multi-robot swarm, a key quantity to be considered is the swarm’s agent density. In the field of multi-robot and multi-agent systems, it has been acknowledged that there is a minimum agent density to ensure the emergence of cooperative behaviors, implying that too few agents within a swarm would yield an ineffective system. However, too large a swarm may result in the agents interfering with each other’s actions, again resulting in subpar swarm performances. There is therefore a range of densities where swarm operations are optimal. In this study, we investigate the factors that determine this range for collective target-tracking tasks. Specifically, we show how the use of agent-based memory can reduce the density at which swarms are able to start tracking. We also show that besides strategy design, other environmental factors affect the range of densities over which swarms can operate efficaciously, such as a target’s movement policy, its velocity, and the number of targets to be tracked.