This note aims to solve a class of consensus problems on switching networks of agents in the sense that all agents reach agreement upon a consensus value but with different time-varying scales. That is, multi-agent networks achieve the so-called "scaled consensus." Necessary and sufficient conditions are established to guarantee the exponential convergence of agents under the action of distributed protocols which are conducted by nearest-neighbor interaction rules. Simulation tests are performed to illustrate the scaled consensus performances for agents in the presence of time-varying scales and switching topologies.