I am regularly asked how to extract the tabular data visible in KML description balloons in ArcGIS Earth into properly typed attributes that can be written to a Geodatabase feature class.
While KML does support a schema object that Data Interoperabilty's Quick Import geoprocessing tool would see and honour, often KML files are received that contain only HTML data in description objects, and the data is frustratingly close but inaccessible.
Data Interoperability is still the answer, but you need to author a Spatial ETL tool tailored to your data. First, inspect the KML features in Earth and decide what attributes you want and what data type they are. Unfortunately there is no magic wand to wave to help with this, the data is untyped in the KML description and you need to do this clerical review.
Usually only one Placemark geometry type is of interest, but if you need to convert multiple Placemark geometry types repeat the review for each one.
Then author a Spatial ETL tool with KML Reader and Geodatabase Writer, manually specifying each output feature class' geometry and attributes.
The data workflow between the Reader and Writer is this:
You may need a GeometryFilter transformer after the Reader to filter Placemarks of interest, and if handling date fields, a DateFormatter to make the values writable to Geodatabase (say %x input, FME date output).
The attributes you expose will then be written as properly typed fields in a Geodatabase.
There is a related workflow to extract attributes from extended data - if anyone needs this message me in GeoNet.