• http://www.urbanbacon.com Phil Novara

    This is great content…once again uncovering the hidden power of Twitter. Google should have analytics updated to accommodate this “hidden traffic” in months to come. I mean, they are Google…

  • http://www.savio.no Eivind Savio

    Great article Danny.
    One other possible reason that may lead to underestimating in Google Analytics as well, is the quality of the traffic from Twitter. If the visitor bounces right back without giving the page enough time to load properly, Google Analytics (and other script based analytics program) may not have enough time to record the visitor. The Google Analytics script is normally placed just before , and this means that the page needs to load before the tracking script is activated.

    I have written a bookmark solution earlier that does the same as Snip-n-Tag, but it works in most browsers:
    http://www.savio.no/blogg/a/82/google-analytics-campaign-url-builder-with-short-url-and-twitter-posting
    (I hope it is OK posting the URL to my own solution)
    It currently only supports Cligs URL shortening, but I may add more shortening services if the demand is there.

  • mayank

    Great analysis Danny. That provides a good data to what all of us have known all along but did not have a convincing proof.

    There however seems to be a small mistake at the begining of the Numbers Bit primarily because of the closeness of the number 58 (bit.ly’s report) and 57 (raw log files). Since you registered the url with the tracking code at bit.ly. The number 58 from bit.ly is most certainly for the url with tracking code. But your log showed only 43 with the tracking code. So there must have been atleast 15 other visits to bit.ly for name lookup, but did not get converted to a click on your site. These would be from twitter clients and browser plugin who would do a url-lookup at bit.ly and show (in the browser/twitter client) the actual url (instead of bit.ly’s). This is most definitely another proof of how overwhelming bit.ly’s reported data is (almost 35%).

    Why can’t we have a protocol for these lookup client’s similar to what robots.txt is for search engine. We could have an optional HTTP header which identifies the request as the one coming from a bot/twitter client/browser plugin and is not expected to be counted. Another option is that since request to shortener’s are only for the link lookup, let them be a HEAD request instead of GET request. A click will definitely be a GET request, but twitter clients can use the HEAD request to only get the expanded URL. I myself use HEAD request for link lookup as it saves some bandwidth by not downloading the entire payload. But there are some URL shorteners that reject the HEAD request (don’t know why), and hence I have to use GET as a fall-back.

    It cannot become a mandatory rule as it will break existing browsers. But robots.txt used by crawlers is also an agreement between crawler and the site. crawlers are not bound to honour it but they do.

  • http://www.brickmarketing.com nickstamoulis

    Wow, this is great info to learn…I am going to start digging a bit more through our log files to try to uncover some additional data. Also, thanks for pointing out the Snip-n-Tag plugin for Firefox as well…

  • http://andrescholten.nl André Scholten

    it looks like mainly non-javascript browsers are not measured. Have you tried to add an extra Google Analytics tracking code that is generated by php, that will measure the non-javascript visits also.

    Dutch explanation about that technique is here: http://andrescholten.nl/google-analytics-zonder-javascript/

  • http://www.ericward.com Eric Ward

    I’d been noticing twitter URLs showing up in my client citation (backlink) analysis data. Since by sheer laziness I have years and years of past client link analysis data, I dumped it all into a seperate directory and searched through literally millions of links looking to see if I could spot any useful backlink analysis trends for tweeted URLs. So far, it’s interesting that twitter permalink URLs (aka individual status update pages) have yet to show up among crawled backlink data, but the main Twitter username URL does show up. In one case from this past May, Google showed 69 unique twitter accounts as having tweeted a particular URL, but none of those 69 were shown to me as twitter permalinks. All were shown as twitter.com/userame, instead of twitter.com/username/status/194816….. Since I can see Twitter does in fact index tweets, and I can also see every URL I ever tweeted, this makes me think tweeted URLs, as they age off the page and become less likely to be clicked, remain silently valueable as quality signal over time.

  • seotoys

    It does look interesting revealing the truth about twitters.. Thanks for uncovering this important fact.

  • http://www.cpcsearch.com Terry Whalen

    Eric,

    When you say tweeted URLs remain valuable as quality signal over time, are you saying that there is an SEO benefit to twitter even though the data shows no twitter permalinks? Would love any additional clarity on this. Thanks!

    Danny, thanks for the post.

  • tigertech

    >(mother’s cookie site:daggle.com was the search, which was me locating
    >the article. Oddly, this request does NOT appear in the raw log files).

    That’s most likely ’cause your computer just served you the page out of its browser cache without contacting the server… but then it ran the JavaScript on the page and thereby logged another hit over at Google.

    – Rob

  • http://borasky-research.net/smart-at-znmeb znmeb

    Well …

    * First of all, no two tools are going to count exactly the same events unless they are designed with the exact same algorithm. In short, you need to approach every vendor / tool maker in the entire analytics space and get their algorithm definition if you want to make sense of the metrics they’re reporting.

    * At the highest level, you probably want to be measuring return (in revenue dollars) per unit of effort: hours spent tweaking the search properties of the web site, time spent in Twitter conversations, etc. The intermediate stuff in the funnel probably matters somewhat, but I think what you really want to know is “should I spend the time optimizing for search, or should I spend the time tweeting, networking on LinkedIn or Facebook, etc.”

    * If the analytics can give you correct *proportions* — what fraction of your incoming unique visitors came from social media and what proportion came from search — that’s probably a lot more valuable than the absolute numbers. And if they can give you paths — do Twitter vistors hit different pages than search visitors — that’s also valuable.

    * I think rather than experiment with the things external to your site, like which tools count what, time is better spent by picking one tool, sticking with it, and doing the experimentation *on* the site. What works best in achieving the overall goals for all visitors, no matter where they came from.

  • http://blog.zebugroup.com Mayank Sharma

    Danny, I think I have come up with a simple solution to remove the counting of bots/spiders/twitter clients/browser plugins from bit.ly’s reported statistics. details are at http://bit.ly/L1ys4 . I am running that experiment on the blog right now and should know the results in a week’s time. Meanwhile would you like to run a similar experiment on your next post?

  • http://www.deskgod.com danhristov

    Danny, I measured many visitors on different social media networks, not only on Twitter. I used web measurement tool RoiWatcher from http://www.Deskgod.com and with this tool i clearly see that no one form these visitors didn’t reach my goals.

  • http://twitter.com/zumbaba zumbaba

    Hi Danny, thanks for this very interesting article though I am still scratching my head to figure out why I have a 7000% discrepancy between Bit.ly and Google Analytics…Does this happen with other URL shorteners like Tiny?