The primordial lithium problem remains a persistent motivation for new-physics modifications of Big Bang nucleosynthesis, yet the precision of the observed deuterium abundance now places strong constraints on such attempts. This indicates that the challenge is not simply to reduce ⁷Li, but to realize the correlated shifts among light-element abundances required to do so without spoiling deuterium. We investigate this issue in a concrete two-step decay scenario involving two unstable particles undergoing sequential late decays. In the first stage, a majoron with lifetime τ_J ∼ 10 -- 10⁴ sec decays predominantly into neutrinos, increasing the neutron abundance and thereby reducing the primordial ⁷Li+⁷Be yield. This mechanism, however, simultaneously drives deuterium above the observationally allowed range. In the second stage, an axion-like particle with a longer lifetime τ_φ≳ 10⁵ sec decays into photons, inducing late-time photodissociation that compensates the excess deuterium without erasing the earlier reduction of lithium, while further amplifying the depletion of ⁷Li+⁷Be. Although the setup is model-dependent, it serves as an explicit proof of concept that the lithium abundance can be lowered consistently with current deuterium constraints. More broadly, our analysis highlights that a viable resolution may require a nontrivial combination of decay channels and decay epochs, and clarifies the pattern of abundance response that successful late-decay scenarios must achieve.