Session-based recommendation, aiming at making the prediction of the user's next item click based on the information in a single session only even in the presence of some random user's behavior, is a complex problem. This complex problem requires a high-capability model of predicting the user's next action. Most (if not all) existing models follow the encoder-predictor paradigm where all studies focus on how to optimize the encoder module extensively in the paradigm but they ignore how to optimize the predictor module. In this paper, we discover the existing critical issue of the low-capability predictor module among existing models. Motivated by this, we propose a novel framework called \emph{\underline{S}ession-based \underline{R}ecommendation with \underline{Pred}ictor \underline{A}dd-\underline{O}n} (SR-PredictAO). In this framework, we propose a high-capability predictor module which could alleviate the effect of random user's behavior for prediction. It is worth mentioning that this framework could be applied to any existing models, which could give opportunities for further optimizing the framework. Extensive experiments on two real benchmark datasets for three state-of-the-art models show that \emph{SR-PredictAO} out-performs the current state-of-the-art model by up to 2.9\% in HR@20 and 2.3\% in MRR@20. More importantly, the improvement is consistent across almost all the existing models on all datasets, which could be regarded as a significant contribution in the field.