Effectiveness of AVO techniques has always been a concern for practicing geoscientists in accessing and modelling reservoirs due to wavelet instability in seismic data. In spite of best efforts made during processing, most often, end product is mixed phase seismic data. This paper is an attempt to study the effect of phase in pre-stack seismic data for quantitative AVO analysis to characterise reservoirs for detection, delineation and development. AVO attributes corresponding to known gaseous reservoirs, computed from zero phase seismic show better AVO anomalies. Prior to AVO analysis, to understand AVO classification, petrophysical analysis was carried out for establishing the relation between elastic parameters and also with seismic signatures. AVO modelling has been used to demonstrate the superiority of zero phase seismic data at known gaseous reservoir sands. In absence of source signature, application of well log data for estimation of wavelet present in seismic was made and by dephasing, the mixed phase data was converted to optimum zero phase. In wildcat exploratory areas, explorationists try to establish AVO anomaly using available mixed phase seismic data. It is found that, at times, AVO attributes are not well defined to differentiate AVO anomalies with reference to background trend extracted from this mixed phase data set. In absence of well data, phase rotation of pre-stack gathers at fixed intervals and their analysis is the best possible solution. This has been discussed in detail by applying on a data set at pre-drill stage and confirming the AVO anomaly which has been proved by subsequent drilling. It is also demonstrated that phase rotation is a robust technique and does not generate spurious AVO response, by applying it at known dry well. This approach of analysing well, synthetic and seismic data together for AVO analysis has increased the level of confidence of interpreters for E & P activities related to the gaseous reservoirs.