Abstract Superposition-time functions offer an effective way for handling variable-rate data. However, these functions can also be biased and misleading. The superposition approach may generally be more useful for well-test analysis (constant rate solutions) than rate-transient analysis. Calculated data points do not tend to be sequential with superposition time but do tend to fall on a straight line corresponding to the superposition function chosen. Examples of superposition are logarithm of time (infinite acting vertical wells) and material balance time (boundary dominated flow). Production data from shale gas wells are usually subjected to operating issues that yield noises and outliers. When the rate data are noisy or contain outliers, distinguishing their effects from common regime will be difficult if the superposition time functions are used as a plotting time function on log-log plots. The superposition function may then lead to a log-log plot that has erroneous straight-line segments. A simple technique is presented to rapidly check whether or not there is data bias on the superposition-time specialized plots. The technique is based on evaluating the superposition time function of each flow regime for the maximum production time. Whatever data are beyond the maximum production time (MPT) are considered as biased data and depend on the superposition function chosen. A workflow involving different diagnostic and filtering techniques is proposed. Different synthetic examples and field examples are used in this study. Once all the problematic issues were detected and filtered out, it was clear that superposition time data beyond the MPT is biased and should be ignored. Thus, the proposed MPT technique can be relied on to detect and filter out biased data points on superposition-time log-log plots. Both raw and filtered data were analyzed using type-curve matching of linear-flow typecurves developed by Wattenbarger et al. (1998) for calculating the original gas in place (OGIP). It has been found that biased data yield a noticeable reduction in OGIP. Such reduction is attributed to the early fictitious onset of boundary dominated flow.