Motivated by the large observed diversity in the properties of extra-galactic extinction by dust, we re-analyse the Cepheid calibration used to infer the local value of the Hubble constant, $H_0$, from Type Ia supernovae. Unlike the SH0ES team, we do not enforce a universal color-luminosity relation to correct the near-IR Cepheid magnitudes. Instead, we focus on a data driven method, where the measured colors of the Cepheids are used to derive a color-luminosity relation for each galaxy individually. We present two different analyses, one based on Wesenheit magnitudes, a common practice in the field that attempts to combine corrections from both extinction and variations in intrinsic colors, resulting in $H_0=66.9pm 2.5$ km/s/Mpc, in agreement with the Planck value. In the second approach, we calibrate using color excesses with respect to derived average intrinsic colors, yielding $H_0=71.8pm 1.6$ km/s/Mpc, a $2.7,sigma$ tension with the value inferred from the cosmic microwave background. Hence, we argue that systematic uncertainties related to the choice of Cepheid color-luminosity calibration method currently inhibits us from measuring $H_0$ to the precision required to claim a substantial tension with Planck data.