For a potential reader there must be some way to go back and see an authentic source you are referring to. I find no meaning in referring to something that only exists in a few *minds*, possibly with different interpretations for the same thing. **Edit*** As mentioned by @mkennedy in the comments below, it is possible to cite *[interview][1]* type communication as personal communication.

One way is to *create* a source, with due permission from the mentioned CEO and then cite it. If possible and everyone attended the meeting agrees; the minutes of the conference may be edited with an addendum stating why it is done so. 
  
  


  [1]: http://www.apastyle.org/learn/faqs/cite-interview.aspx