I launched this blog yesterday and the only promotion that I did was to my Twitter followers and Facebook. I have around 20,000 followers currently and approximately 550 friends on Facebook. I promoted the blog post really late at night on the 26th and received approximately 26 visits that night with another 159 visits the next morning. Granted, I should not write blog posts late at night if my only promotion method is via Twitter as most people are asleep and probably have their updates turned off. But that’s another blog post.
What I discovered is that I have 102 visits that came directly to my blog post, 54 visits from Twitter.com, 21 from GapersBlock.com (a large Chicago blog that was kind enough to link to the post), and 15 referrals from Facebook.
Now I am not the most logical individual sometimes, so feel free to poke holes in my logic in the comments, but it seems to me that nearly 100% of the direct referrals are probably referrals from Twitter applications. If this is true then the math looks like this . . .
102 direct referrals / (102 direct referrals + 54 Twitter referrals) = 65% untracked Twitter referrals!
If that is the case then there are a TON of companies out there that are making the wrong decisions about Twitter if they are basing those decisions off of their analytics data.
Whoa, that’s a massive miss ratio. A couple questions arise in my mind…
- How much of that is uncategorized spidering?
- How could we go about verifying this phenomena, and getting better numbers on it?
- What about url shorteners? Could those be used as a clue somehow?
I think there’s two ways to nail this down: the first is by counting noses as far as the twitter apps go, and averaging to the mean; the second way is to get a good sample space of naturally popular twitteres to make the same test you have.
There’s problems with both approaches… but they’re both doable with a bit of work. Why can’t we set up a site with, say, 1000 unique images and see if there’s at least 50-100 heavy twitter hitters up for the idea of passing the urls around to see where their followers are coming from?
I’ve been following your tweets for some time, by the way. Entertaining and informative. Regards, Don (@Parahacker)
Don – Omniture doesn’t count spider traffic in their stats (only server log files catch those). Also a URL shortener (if it 301s, some don’t) would show the originating URL to Omniture.
I’m not sure I follow you on the test though. Could you expand upon that?