In this paper we argue that having available higher order fully exclusive Monte Carlo generator for Higgs boson production from heavy quark fusion will be mandatory for data analysis at LHC. The H → τ τ channel, a key for early discovery of the Higgs boson in the MSSM scenario, is discussed. With simplified example and for m H = 120 GeV we show, that depending on choice among presently available approaches, used for simulation of Higgs boson production from bbH Yukawa coupling, final acceptance for the signal events being reconstructed inside mass window may differ by a factor of 3. The spread is even larger (up to a factor of 10) for other production mechanisms (promising for some regions of the MSSM parameter space). The complete analysis, which necessarily will add stringent requirements for background rejection (such as identification of b-jet or veto on b-jet) and which will require statistical combination of samples selected with different selection criteria may only enhance the uncertainty.