The paper presents an interesting attempt to estimate CCN concentrations from lidar observations with the support of ground-based in situ CCN observations.

I have only a few point. Minor revisions are needed.

Detailed comments:

P7, l218: kNN is not explained in the main text body, only in the abstract. Thus, please explain the abbreviation here, i.e., mention: k-Nearest Neighbors classifier

P5, Data section

In this section, one should present a simple sketch that provide an overview of the entire data analysis concept. This sketch should include the near-surface in-situ CCN observation (station), a lidar backscatter profile (from the ground up to a cloud base…), a horizontal line indicating the top of the aerosol layer (boundary layer) and a tiny cumulus cloud with cloud base height close to or at the boundary layer top (and the lidar profile ends at cloud base), and maybe even another  (varying) vertical line indicating homogeneous aerosol type conditions or slightly varying aerosol-type conditions with height …… In this way the reader would be able to easily understand the entire retrieval concept.

P18, l491: The method is exclusively applicable to PBL clouds? What about CCN estimates in the free troposphere. Should be discussed.

My comments are located in the attached document.