Caveat to tracking short URLs
The rise of Twitter has created a huge need for creating shortened URLs to fit within the 140 character limitation. Additionally, we've also witnessed an increased desire to measure the effectiveness of those clicks through shortener tools like Bit.ly, Ow.ly, and Argyle. All of these tools do a good job of collecting click data and offering other insights into behavior, but there are some things to be aware of before you blindly use the data you see in these tools.
One of the biggest caveats for the shortener tools are the click discrepancies you'll notice between your favorite web analytics solution (Google Analytics, WebTrends, Omniture, etc) and what the shortener tool tells you. Why is that? All tools collect and process data a little differently and will always have a difference in the numbers. URL shorteners have an even bigger problem as most (if not all) don't filter out robots and spiders that are crawling links. Essentially, they are counting non-human clicks.
In Twitter, this would add only a handful of additional clicks to your totals. However, the effect becomes intensified if you happen to post the short URL on a web page, especially a blog which is constantly being crawled. Blog comment sections are notoriously crawled by spambots and as a result create artificial clicks to your nice little short URL. We recently found this out in a short URL we created which showed 860 clicks in Bit.ly but only around 80 in Google Analytics. We traced it back to some external blog comments where we posted a link to our short URL. Apparently the spambots loved our short URL!
At this moment I am not sure if there is a workaround but I would imagine over time these tools will get smarter about filtering out the inhuman traffic. They'll have to in order to become useful. My advice is to use the data with high confidence if you are using it to only track Twitter and other social media, but with increased skepticism if you are posting links on websites using the short URL. If you want to measure real outcomes it is always a best practice to embed tracking code parameters to be picked up by your analytics platform as they filter out the bots. We'll dig into that in more detail at the Social Media Bootcamp next week.
Comments
Well said, Jim. Unfortunately there is also a pretty high incidence of bot clicks for links posted to Twitter. Based on our analysis thus far, it looks like 20-30 per day. My hunch is that many of these are spam detection mechanisms and the rest are content aggregators.
There are ways to weed out these clicks. We're working on it. :)
Eric @ Argyle
See Eric - I knew if I linked to you I'd get excellent commentary. Mission accomplished.
That is interesting that the bots are higher than I thought even for Twitter. I was thinking it would only be a handful.
So scratch what I said above...just use your web analytics platform (where you can) to track outcomes until you guys create a robust filter that will solve world hunger. Will that be ready next week? :)
The click numbers for in-stream links may not be completely accurate, but they are consistent. So you can still use the data to make comparisons, built aggregates, etc.
The "more accurate click counts" project is on the list...but not for next week. :)
Leave a Comment