Posted April 29, 2009 11:12 am by with 30 comments

Tweet about this on TwitterShare on LinkedInShare on Google+Share on FacebookBuffer this page

Nielsen Wire posted a shocking statistic to its blog:

Currently, more than 60 percent of Twitter users fail to return the following month, or in other words, Twitter’s audience retention rate, or the percentage of a given month’s users who come back the following month, is currently about 40 percent.

The post also included a comparison of Twitter’s retention rate, compared to Facebook’s and MySpace’s:

But–and it’s a big but–did Nielsen fail to take into account that many Twitter users start off using the web site interface, then quickly migrate to a third-party application? That’s the suggestion Brendan O’Connell offered up.

A quick look at Twitstat seems to back up his theory. Twitstat is tracking over 200 different Twitter applications and, as the chart below shows, only 27% of Twitter users are using the web interface:

Of course, if we’re to question Nielsen’s numbers, we should also question Twitstat’s, but that 40% retention rate may not tell the full story.

UPDATE: For those of you that assumed Nielsen would have been smart enough to include external apps, it turns out the company did forget. It also turns out that the retention rate was still only 40%.

  • Andy,

    Good post. This is exactly what I was thinking this morning. Is the measurement, or twitter the service.

    Mike Lizun

  • Good point, however I am a typical twitter user with large following @theReviewGuy, based on my experiences, Neilson’s analysis seems on track. I use all media while twittering, and I doubt as a leading research agency, Neilson did not think of the multiple access points as a characteristic of twitter usage. I like twitstats, but it’s a rather simple application that simply pull API data, not much human analysis behind it.

    Grant’s last blog post..Twitter Quitters Post Roadblock to Long-Term Growth

  • Mer

    Just the fact that there are over 200 Twitter user apps is telling. I chose the Twitter client Twitterific because it was designed for Mac users, and therefore automatically friendlier for me than a typical website. It was the first thing I sought out once I began daily Twitter use. Stats are often misleading and should be very cautiously applied.

  • I am sure if I am advertiser I would like to know the difference…WHEN twitter starts offering advertising that is.

    Jaan Kanellis’s last blog post..Brain to Text Has Arrived For Twitter [Electroencephalogram]

  • Good catch! And I agree with Jaan – knowing the difference could really impact future advertising revenue.

    Data points, Barbara

  • Looks like Myspace is still going strong, still dont think it was worth what they paid for it though, anyone know how much they are making off it?

  • Stu

    I thought something smelled a little funny when I read that stat.

    Thanks for bringing this up, it makes a lot more sense than the Nielsen stats.

  • Sorry to disagree but three points –
    1. As many said, the statistical methods that Nielsen uses would have accounted for any multi access points. This would be a stats 101 mistake and they are too experienced for that –
    However even if they did –
    2. Even if they only pulled web data all 3rd party apps pull API from Twitter itself and post to our pages, so let’s say Nielsen made that error of not accounting for multi-data points (v unlikely) it would not have mattered – But let’s say all these are wrong
    3. My guess would be that Nielsen and HitWise work together and HitWise has actual IP log data – so even if their methods were wrong and somehow the fact that are usage all appears on our pages was mistakenly analyzed – Hitwise would be able to give very good actual data…

    I do think that the need to download apps to really work with Twitter, that most people do not own iPhones and that many people are barely tech savvy or have little time to care about more than email or a few texts would make Twitter more of a hassle to many who start.. Just my opinion, but I believe it is very possible retention rates are that low.. Much of how to use Twitter is a mystery, they offer no explanations on your pages and people do not like to feel lost. My guess is most people feel lost and give up.

  • Oh forgot something..
    Sorry – but the other app displayed here with stats is irrelevant because you would have to assume one technology for posting. Many people use multiple types, so there is nothing you can deduce from that in a comparative or accumulative model. All you can do is say there are x people who use y product. Nothing else. As a statistical application Twitstat shown here, has no applicability to the discussion. They are not valid as a comparative/cumulative tool for analysis.

  • What i liked about the statistics is that myspace still keeps its positions but look how facebook is growing, step by step.

  • @Kristine – but how would Nielsen know how many people are users of a particular 3rd party app that pulls an API to a dashboard client? It doesn’t even involve a web-based session?

  • The tools I use push and pull date from the API which gets it data from the same DB as the website, then the posts all go to the website. I do not know of an app, though I do not know all, so could be wrong, that does not post your 3rd party API tweets to your page.

    My client takes in the data and pushes it out to the site where it sits. You can see how many updates I made, what they were, my DM’s, my @schachin. All on my page. The app is a push pull to the site and from the site in an XML feed that is known as the API (well typically XML).

    That means there is a record on the website of the activity. They may be using log data from hitwise or they may have other methods to monitor that activity (API?), but the data is all on the site to be seen and consumed on page or through API’s. NO data is lost between your client and the page. The API just pulls and pushes information back and forth between client and site.

  • Oh and sorry 🙂 meant to say hi first! 😉
    Hi Andy!
    Ok so works like this >

    You post to Twitter on page.
    TwitterSite –> TwitterDB —> TwitterAPI —> ClientAPP

    Then you answer from your ClientAPP – which then goes
    ClientAPP –> TwitterAPI –> TwitterDB –> TwitterSite

    Same API, Same DB – same data source- same activity as if you posted on page.
    IE What you see in the app, you see on the page.
    At least that is the usual path in a push / pull API system.

  • @Kristine – I guess without confirmation from Nielsen on how it came up with its numbers, we’re left making assumptions about what it can and cannot track. 🙁

  • Kristine Schachinger @schachin

    Somewhat, but not really. I think my point was lost in this. What your post said was they were wrong because they could not count 3rd party app tweets, so data was missing therefore there conclusion was wrong. Correct interpretation, yes?

    What my post said is not how they did it, but no matter how they measured the app tweets made no difference because the data was not missing. The data was available just as any other tweet data was available. However, the retrieved the app data from the site it all came from the same DB and all was available on the site. Therefore no data was missing.

    How they measured would confirm they knew how to measure data in the first place, but there was not data missing, so not really an effect in this debate.

  • You have a valid data point. One should define a parameter and then measure the retention. eg no. of tweets in x no. of days. this should offer a better data point and also eliminate the data deviation that is there in Nelsen report due to multiple twitter clients.

    Samraat Kakkar’s last blog post..The Twitter Burst In India

  • Hey Andy-

    Thanks for starting this discussion. Would be nice if Nielsen would reveal their methods, but they have been around a while and usually do a pretty good job gathering their stats. Either way Twitter has a technical challenge of scalability which must be overcome, in doing so, they will also overcome the social perception that there service is unreliable. Lately with the rush of interest, the service has been frustrating for many.

    Travis Campbell’s last blog post..Your Business Plan for Failing Social Networking Sites

  • What, you’ll trust them to measure search market share but not to measure Twitter user engagement?

    Tsk, tsk, tsk.

    Michael Martinez’s last blog post..Matt Cutts stresses Google crawling and indexing

  • Be sure to read the update before adding a comment. It appears Nielsen didn’t take into account external apps. It also turns out it doesn’t effect the outcome.

  • Pingback: Twitter Today: May 1, 2009 | kenneth lim . net()

  • A big deal was made when Oprah got a twitter account. Despite not having posted anything in the last 5 days on Twitter, she is still gaining hundreds/thousands of new followers every day. To me, this says more about the power of her image and television has on the media-obsessed public than on the success of Twitter. But this pattern is typical, a spike in usage, followed by a descending drop, sometimes very precipitous.

  • Pingback: How Loyal Are Twitter Users? | AccuraCast Search Daily News()

  • Pingback: Dear Twitter - Congratulations! |

  • Wait until the next big thing comes out and watch everyones stats drop off the charts.

  • Pingback: Is Twitter Dead?()

  • Nice article! you don’t need to explain more as the graphs says everything.

  • Twitter retention is very low, and it is very obvious.

    Everyone knows the majority of “join requests” come from spammers who are building their follow list. The few real “join requests”, we get are easy to review and we can their activity. When I review my “joiners” a month later, it is obvious that a most of them have become “dead accounts”.

    In my opinion, the most important reason for people leaving Twitter, is that the conversations are shallow or weak. Most “tweets” are statements, one-directional comments, insignificant rants. Insufficient “tweets are actual conversations. So the members do not feel they are part of a community, there is little to draw them back.

  • Pingback: Nielsen Could Have Twitter’s Retention Rates Wrong | Move with Mind()

  • Pingback: Keeping Tweeters on Twitter « Social Media Snippets()

  • Pingback: How Loyal Are Twitter Users? - AccuraCast()