<p>Synchronization in complex dynamical networks of nonlinear systems is a fundamental problem in various scientific and engineering domains and has been widely studied. However, achieving synchronization in cluster based networks with coupling limited to output states, particularly using contraction theory, remains a significant challenge and has been only minimally explored. This paper addresses a generalized strategy for output feedback based synchronization of cluster based networks using contraction theory. The proposed framework ensures global exponential synchronization by establishing explicit conditions for controller gains in diffusively coupled nodes. These conditions are derived by computing upper bounds on the respective states, obtained through eigenvalue analysis. To validate the effectiveness of the proposed approach, cluster based network synchronization results are demonstrated on a cluster based complex dynamical network featuring three different scenarios: (i) an all-to-all network of coupled Lorenz systems, (ii) an arbitrarily connected network of Rossler systems and (iii) all-to-all connected network with heterogeneous clusters. The proposed contribution can easily accommodate the scenarios where each cluster has to synchronize to a separate manifold. Theoretical findings are rigorously confirmed through comprehensive numerical analysis, demonstrating the efficacy and applicability of the developed synchronization technique.</p>