Link shorteners like bit.ly, goo.gl, and t.co generally keep their own metrics for how many times users click through. Unfortunately, those numbers rarely match any other Analytic’s system, including Adobe Analytics. We’ve seen situations where the bit.ly number was as much as 10 times the analytics number. With such a big difference, both numbers look a little suspect. So, what gives?
First, there’s non-human traffic. Link shorteners’ numbers do not always account for whether the click came from a real live human being, or a bot. Goo.gle and bit.ly DO try to eliminate bot traffic, but the list they use to identify bots is likely different from the list Adobe uses (which is maintained by the IAB).
However, link shorteners are often used on twitter, and bot traffic on twitter is a whole different beast. Web bots generally crawl pages, sometimes searching for specific things. Adobe and the IAB keep a good list of the worst web-crawling bot offenders. But on twitter, the environment is more predictable (tweets, unlike webpages, can predictably be accessed/interacted with the same methods) and the actions bots perform are simpler: click, retweet, reply. This means many more “hackers” out there can create a bot with much less effort, and since they aren’t harming much, they fly under the radar of the IAB and other bot-tracking organizations.
I found a book, “TWITTER: The Dark Side – Does Bit.ly Enable a Massive Click Fraud?” (by Roman Latkovic and Robert LaQuay, Ph.D) which takes a thorough look at this issue. Basically, when you start to break down individual clicks by IP and user agent, it becomes clear many bots are indeed inflating bit.ly tracking.
Lost Referrer Information
Third, bit.ly (or twitter traffic in general) may cause you to lose referrer information. Someone coming from a twitter app rather than a webpage won’t have a referrer-it may look like “Direct Traffic”. Query string parameters SHOULD always carry through, though, so make sure your shortened links include tracking codes.
With this information in mind, I’d be more surprised to see bit.ly numbers that MATCH an Analytics tool than I am to see wide discrepancies. What’s your experience, and what are your theories behind them?