A new paradigm is emerging in the field of photochemistry modeling in giant planets and Titan atmospheres. The emphasis is placed on the accurate predictions of the models and the quantification of their uncertainties. In order to improve photochemical models pre- dictions, it is necessary to identify in chemical schemes the key reactions that should be studied in priority at conditions relevant to plan- etary atmospheres. We discuss the notion of key reactions in chemical schemes and we compare several methods dedicated to the search for these reactions. Our study focuses on the methods that have been published in the field of hydrocarbons photochemistry for giant planets and Titan atmospheres. These methods can be grouped into three families: determination of production paths, local sensitivity analysis methods and global sensitivity analysis methods. We first compare all these methods using a simple 0D photochemical toy model of Titan's atmosphere as a study case. Then, we applied sensitivity analysis methods to a realistic hydrocarbons scheme. We show that, due to the current large uncertainty factors of the rate constants at low temperature (100 K < T < 200 K) and the nonlinearities in hydrocarbons photochemical models, global methods should be preferred to infer key reactions.