Hybrid organic-inorganic lead halide perovskites have emerged as promising materials for solar cells, attracting significant research interest over the past decade. A key requirement for fabricating efficient and stable perovskite solar cells (PSCs) is to minimise non-radiative recombination losses, such as trap-mediated and interfacial recombination, so that, under open-circuit conditions, recombination is dominated by the radiative channel.^[1] To monitor the charge transfer dynamics in perovskites, transient spectroscopy techniques, such as transient photoluminescence (TRPL), are widely employed. However, the analysis of the TRPL decays becomes increasingly complex as additional layers are deposited on top of the perovskite. For example, passivating interlayers or charge transport layers can introduce additional midgap states or give rise to asymmetric trapping, further complicating the physical processes taking place following illumination. Moreover, the quantitative analysis of TRPL traces is currently performed by applying biexponential fitting functions or the ABC model, both of which fail to capture the full physical picture in the nanosecond-to-microsecond time regime, as well as the complicated recombination processes governing the perovskite bulk and its interfaces.[^2,3] It is therefore crucial to ascribe physical meaning to the observed transients in order to properly describe the underlying dynamics in a quantitative and analytical manner. ^[4]

Herein, we propose combining multi-parameter recombination models ^[5] with a Bayesian inference framework ^[6] to analyse TRPL decays in perovskite thin films and half-stacks with solar cell-relevant architecture. The Bayesian framework, built on a Markov-Chain Monte-Carlo (MCMC) sampler, explores the highly correlated, multi-dimensional parameter space of the proposed physical model and is applied directly to experimental data. We show that incorporating multimodal data into the Bayesian inference pipeline, specifically nanosecond transient absorption spectroscopy (nsTAS) and photoluminescence quantum yield (PLQY), along with TRPL, better constrains the parameter space and yields a more accurate quantitative description. We further apply this methodology to elucidate the effect of various interlayers on recombination dynamics following photoexcitation, distinguishing between chemical and field-effect passivation mechanisms, and between non-radiative recombination and charge extraction. This establishes a robust protocol for systematically guiding future material and device development.