Inflation is the leading theory to describe elegantly the initial conditions that led to structure formation in our Universe. In this paper, we present a novel phenomenological fit to the Planck, WMAP polarization (WP) and the BICEP2 data sets using an alternative parametrization. Instead of starting from inflationary potentials and computing the inflationary observables, we use a phenomenological parametrization due to Mukhanov, describing inflation by an effective equation of state, in terms of the number of e-folds and two phenomenological parameters alpha and beta. Within such a parametrization, which captures the different inflationary models in a model-independent way, the values of the scalar spectral index n(s), its running and the tensor-to-scalar ratio r are predicted, given a set of parameters (alpha, beta). We perform a Markov Chain Monte Carlo analysis of these parameters, and we show that the combined analysis of Planck and WP data favors the Starobinsky and Higgs inflation scenarios. Assuming that the BICEP2 signal is not entirely due to foregrounds, the addition of this last data set prefers instead the phi(2) chaotic models. The constraint we get from Planck and WP data alone on the derived tensor-to-scalar ratio is r < 0.18 at 95% C.L., value which is consistent with the one quoted from the BICEP2 Collaboration analysis, r = 0.16(-0.05)(+0-06), after foreground subtraction. This is not necessarily at odds with the 2 sigma tension found between Planck and BICEP2 measurements when analyzing data in terms of the usual n(s) and r parameters, given that the parametrization used here, for the preferred value n(s) similar or equal to 0.96, allows only for a restricted parameter space in the usual (n(s), r) plane.