Tony Karrer's eLearning Blog on e-Learning Trends eLearning 2.0 Personal Learning Informal Learning eLearning Design Authoring Tools Rapid e-Learning Tools Blended e-Learning e-Learning Tools Learning Management Systems (LMS) e-Learning ROI and Metrics

Tuesday, September 08, 2009

Twitter Brings Lower Quality Clicks

Richard Hoeg points out that for his blog Twitter = High Visits But Low Conversion. Basically, he shows his "referring sites" from Google Analytics for the past two weeks:


His conclusion:

Folks who visit from Twitter don't visit as many pages and spend less time of the site.

Of course, that made me wonder if twitter really was bringing lower quality clicks than other sources. That's contrary to what I would expect. You would think that someone who gets a link referred by someone they know would visit and then look around. It should be pretty qualified. So, I looked at a similar view of referring sites:


Indeed, people coming from twitter are the lowest in pages viewed per visit and near the bottom in time on site and highest bounce rate. Likely they were interested in the specific item that they came there for, but still it's a bit disappointing that they don't click around a bit more.

Of course, a relatively small percentage of traffic from twitter actually comes from "" – many people use tools like TweetDeck. And I believe many of these are reported as Direct Traffic. So, I went to the list of All Traffic sources:


and while Direct Traffic does have a good number of referrals, it really doesn't provide good results. Basically, it's about the same as organic search traffic. And some of the Direct Traffic that comes from twitter is lumped in with Direct Traffic from other sources including RSS readers. And I believe that those other sources likely are higher quality clicks. Notice that as a referring source (likely Google Reader) is better than Direct generally. Bloglines also has better numbers.

I tried to get a bit more detail by using to see more about sources, but unfortunately, they also run into the same issue with the different twitter sources. Here's their description of "referring sites":

Direct Traffic includes people clicking a link from:
- Desktop email clients like Microsoft Outlook or Apple mail
- AIR applications like Twirhl
- Mobile apps like Twitterific or Blackberry Mail
- Chat apps like AIM
- SMS/MMS messages
It also includes people who typed a link directly into their browser

So they can't help differentiate either. Bottom line, everything I'm seeing suggests that Richard was right:

Twitter brings lower quality clicks

What's also interesting here is that there's been quite a bit of high profile discussion around Does This Blog Get More Traffic From Google or Twitter? where there was a question of whether twitter brought more traffic than traditional sources. For Fred Wilson, he gets pretty huge twitter traffic.

For Richard and I, we don't get nearly the same levels and it's not even close.

Twitter delivers some traffic, but it's still small compared to search.

It's surprising that Fred Wilson is not looking at the question of the quality of his twitter traffic either.

Aggregators Bring Traffic

One last thought, it's been a while since 2007 Traffic Stats - Hopefully a Meme where I looked a bit at my traffic numbers. They've grown considerably over the years, but a lot of the statistics have remained consistent. One of the really interesting things I saw in Richard's stats and in my stats was:

Two Aggregators (eLearning Learning and Work Literacy) are among the top 5 in referring sites.

On Richard's eLearning Learning was number 7 as a referrer. For him, they were 100% new visitors and had pretty good pages clicked and time on site. For me, it was also pretty good quality traffic.

This is somewhat validating the concept behind these sites and the Browse My Stuff concept.

And all of this makes me think:

Marketers interested in quality clicks should focus less on twitter and more on blogging, search and aggregation.


Sreya Dutta said...


I use ice rocket to monitor my blog and i have the exact same experience as you and Richard. It is mostly and direct that gives the most results!


V Yonkers said...

Tony, since you brought it up, I was wondering, what is a bounce rate? I have seen these on both the statcounter and google analytics.

I understand the other information, but I think you could break down the information more to determine why there is a more "shallow" click using twitter. For example, your top posts may have been twittered or they may have key words (I find "define" in a title tends to make that post a top post). Did all the twitter come from the same region? What is the native language of the twitter post readers? Is it possible that they expected more graphics? I'd be interested to see what a deeper analysis could tell you about those that come from a tweet.

Tony Karrer said...

@Sreya - I use icerocket as well to scan for posts that reference my various sites. I also use BackTweets to look for tweets that reference my sites.

@Virginia - A "bounce" is someone who comes and looks at only one page. In other words, they take no additional action.

I've taken a quick look at things like regions, etc. and don't see a whole lot to suggest why they are more shallow. Some of the other things such as graphics, etc. is the same when they come from other sites (e.g., a blog reference) or via search.

I'm going to definitely keep my eye open to see if I can tell more about what's going on.

Stephen Downes said...

So, the 'quality' of a click is based on how many pages they visit and how long they stay?

For me, a high quality click is one where the person lands on the page they want right away, gets what they need, and gets back to their work with minimum fuss and bother.

Sam said...

Hi Tony,

Good post. I have had a similar experience as a marketer promoting white papers, webinars, etc. The absolute clickthrough #s from Twitter, et al seem very encouraging, but the conversions are far below what we would normally get with email or another method. I have had abandon rates of 95%+.

With Twitter traffic, it is easy come, easy go I suppose.

Tony Karrer said...

@Stephen - there are certainly cases when people will come looking for something specific and leave as soon as they find it. I use social signals other than time and number of pages to determine value of content.

That said - if someone comes and quickly leaves, chances are they didn't find much value.

@Sam - interesting comment. I'd think that someone who's subscribed has already said "that content is interesting" so that might not be a fair comparison. However, search traffic vs. twitter vs. linkedin vs. other new visitors is fair to compare. And it's surprising that twitter would not do as well as search - at least it was to me.