I was charged with getting the number of times a specific PDF has been clicked on our site for the past two years. GA registered something like 30 downloads, which at first I found impossible to believe but it is an obscure academic report so I said OK. Person connected to the report couldn't believe it, she said that she has seen sites where this PDF had 8,000 or so downloads over same period. Well Webtrends for our site showed that over 16,000 downloads to that one URL during that same period and 8,000 unique visits.
Now I know that server log statistics are practically useless and that when a user clicks on a PDF, the log file can register a bunch of requests to the server but still, this discrepancy is astounding. Even after I filtered out 206 HTTP status codes and as many bots as I could, I still had about 8,000 downloads with about 6,000 visits.
We do use tracking codes, but I am using unique page views for this in GA. Tracking codes actually bring up even fewer downloads. Am I right in concluding that a tracking code won't track a PDF linked directly from a search engine, either Google or any other - because obviously our tracking code isn't on any of their pages. Much of our content comes from Google searches. I cannot find keywords or phrases that would reasonably correspond to this PDF in analytics for search terms either.
So now I have no idea what to tell this woman. I tend to trust GA WRT pdf downloads. I also find it hard to believe there were only 30 downloads in two years - even an obscure academic PDF. I need to figure this out. Any suggestions?
Any advice is much, much, appreciated.