Massive stars ($\ge 8\, \rm {M}_{\odot }$) play a key role in shaping the interstellar medium of galaxies through stellar feedback. However, how these stars form and evolve before exploding as core-collapse supernovae (SNe) remains elusive. We compute for the first time the star formation efficiencies (SFEs) at the locations of hydrogen-rich (H-rich) SNe and stripped-envelope SNe (SESNe) to constrain their progenitor properties. We used VLT/MUSE and ALMA observations of $\mathrm{H}~\alpha /\mathrm{H}~\beta$ and CO(2--1) emission lines to trace the components of the warm ionized gas and cold molecular gas, respectively. Both observations resolve individual ${\rm H\, {\small II}}$ regions and giant molecular clouds at spatial resolutions on cloud scales ($\sim 100$ pc). These combined data allow us to compute the SFE from the star formation rate (SFR) and the molecular gas mass ($M_{\rm mol}$) as $\mathrm{SFE}=\mathrm{SFR}/M_{\rm mol}$. We find that SESNe explode in environments that are currently forming stars eight times more efficiently than those of H-rich SNe (higher SFR for SESNe with similar $M_{\rm mol}$). On one hand, this is consistent with the scenario in which the majority of SESNe are produced from very massive stars ($\ge 20~\rm {M}_{\odot }$) if the initial mass function is top-heavy. On the other hand, most of SESN progenitor channels are formed from interacting binaries ($\le 20~\rm {M}_{\odot }$) if an increased binary system formation rate is connected with turbulences and, in turn, with the boost to SFE. Then, an increased binary fraction could explain the enhanced H $\alpha$ luminosities. In summary, SESNe preferentially occur in regions of intense, efficient star formation rather than simply higher gas content.