Transposons, which are DNA sequences that can move to new positions in the genome, make up a large fraction of eukaryotic genomes and occur in clusters. The insertion of transposons into the genome is hindered by compact folding of chromatin, supposedly preventing aberrant or even pathogenic insertion. Chromatin can, however, be decompacted as a consequence of transposon insertion, leading to increased accessibility and, in consequence, further insertions. While these observations suggest a positive feedback between chromatin unfolding and transposon insertion, how such a feedback might contribute to clustered transposon insertion remains poorly understood. In this study, we analyze polymer models of a self-interacting chromatin domain that unfolds as increasing numbers of transposons are inserted and block the self-interaction. On the one hand, we find that, if additional transposons are inserted adjacently to already inserted transposons, the unfolding of the chromatin domain changes from a sharp globule-coil transition to a more gradual extension of loops from a core that remains folded. On the other hand, we find that adjacent transposon insertion emerges either when transposases are excluded from densely packed chromatin, or when transposon insertion proceeds very quickly in relation to the thermal equilibration of polymer configurations. We thus derive from our model physical conditions for clustered transposon insertion and the resulting spatial compartmentalization of chromatin. An according role was recently suggested for LINE-1 and Alu repeats, which occur in clusters and drive the mesoscopic compartmentalization of the mammalian genome.