Since Amazon uses the borrows in sales ranking, it obviously knows what the number of borrows is and should give us that information. I can estimate approximately how many borrows I might have had based on pages read, but it would be easier to estimate audience size if I didn't have to do that.
So true. The deliberate opacity of data from Amazon is a constant source of frustration. Because of that, when I'm guestimating the number of KU downloads I operate from the assumption that everyone who downloads the book reads the whole thing. I know that's utter BS, and I know it likely under estimates the number of downloads by a wide margin... but it keeps things simple.
I think most people who try to estimate their audience size use FREs (full-read equivalents) just as you--because how else would you do it? That probably does underestimate the number of downloads, but in KU, one can get a download that never results in any pages read, and it's arguable whether that should count, anyway. However, we don't know whether buyers read the book, either. It's logical to assume most of them do, but I know I sometimes buy things, intending to read them, and then either don't or don't for a long time. For that matter, some of the returns (which I don't don't count as a sale for audience purposes) might be people who read the book and then returned it.
We can't really tell how reads equate to buys or borrows, so it would be easier to have consistent data for each. Number of buys and number of borrows makes sense.
