In this work, the origin of singlet and triplet exciton-induced degradation of host materials with C(sp2)N(sp3) bonds around nitrogen (carbazoles, acridines etc), connecting donor and acceptor units, were unravelled using DFT and CASSCF methods. The results reveal that molecules (employed in OLEDs) with basic unit containing C(sp2)N(sp3) bonds (nitrogen connected to carbon in triangular fashion) have natural tendency to fragment at C-N bond through S1/S0 conical intersection (CI). The calculation of barrier heights, to reach dissociation point, indicates that degradation via triplet states is kinetically less feasible (∆G*T1-TS >25 kcal mol-1) compared to first singlet excited state (∆G*S1-TS 7-30 kcal mol-1). However, long lifetime of triplets (as compared to singlets) aids in the reverse intersystem crossing from triplet to singlet state for subsequent degradation. From the results and inference, ∆G*S1-TS and ES1-T1 are proposed to be the controlling factors for exciton-induced degradation of host materials with C(sp2)N(sp3) bonds. Further, multiple functionalization of carbazole moieties reveal that polycyclic aromatic systems employed as acceptor unit of host materials are best suited for PhOLEDs as they will increase their lifetime due to larger ∆G*S1-TS and ES1-T1. For TADF-based devices, materials with fused ring systems (with N(sp3) at the centre) in the donor unit is the most recommended one based on the findings of this work, as it avoids the dissociative channel altogether. Negative linear correlation between ∆G*S1-TS and HOMO-LUMO gap is observed, which provides indirect way to predict the kinetic stability of these materials in excitonic states. These initial results are promising for future development of QSAR-type approach for smart designing of host materials for long-life blue OLEDs.