7 Reasons Not To Use The New Tweet Buttons

1. Nofollow

The Twitter retweet code places a link to Twitter on every page you include it, and they didn’t add a nofollow to the link. It is quite possible Google will decide to ignore all these links in the future, especially as it is effectively hidden and not a “vote” for a particular page. The destination of the link doesn’t show similar information to what is in the button.

2. History

The button has no historical count of tweets as is clearly evident if you search for tweets on something “historical” such as an individual domain. If I added a Tweet button to my home page the link through to search for Tweets would be something like this.


3. Count Accuracy

If I glance back at Shaun’s post, the tweet button currently says 23 Tweets – clicking the button to see those tweets results in only 17 tweets showing.

Here are just the influential Tweets counted by Topsy, half of which Twitter hasn’t found even though they actually use the button to submit their Tweet.
Tweets on Topsy

Here is a link for the Twitter search results to see if they ever catch up with Topsy.

4. Longevity

Lets look at those links for a second



Topsy uses a permalink – there will be a permanent record of all of those Tweets, and their history goes back a long way, and even copes with 301 redirects to some extent if you change your permalink structure. I haven’t experimented with this extensively, but that has some useful potential for marketers adding parameters to links.

For Twitter the link is one of their funky search URLs which uses a # named ancor that is ignored by search engines – Twitter search is totally unreliable, and only has a 7 day history – do you think you are going to have a record 1 month, or 1 year down the line?

5. Broken Retweets

The new “expanded” links on the web interface get truncated – if you retweet using copy/paste links will quite often end up being broken.
This is a subtle way of enforcing retweets using their “new” official retweet method which still isn’t as popular as Twitter would like people to believe.
It also defeats the purpose… you can’t see the whole URL, sneaky redirects or affiliate tracking parameters, and for many URLs might even hide what the URL is actually about, or where you might end up.

6. Validation

Shaun is pretty hot on W3C validation of websites though maintaining it on content that is being published in response to news is a challenge.
One of the problems with the original Tweetmeme button was validation so I am hardly being a validation nazi here.
http://validator.w3.org/check?uri=http%3A%2F%2Fwww.hobo-web.co.uk%2Fse o-blog%2Findex.php%2Fnew-twitter-button-conveniently-without-nofollow- attribute%2F&charset=%28detect+automatically%29&doctype=Inline&group=0

Twitter decided to invent their own link attribute – that sucks – they will never manage to get that accepted as valid code, as it is proprietary to them and doesn’t serve any purpose. This was one of the major complaints against rel=”nofollow” which took 4 years to get any kind of official acceptance.

7. Data

He who has most data in online marketing often has a huge advantage. By adding the official Twitter button they suddenly gain a whole load more information about who visits which sites, which can be tied into their member profiles.
I can’t currently see a reciprocal benefit in Twitter having access to that data.

I gave Twitter a 7 day Ultimatium – now 6 months ago
I strongly criticized their search and the ability to find my own content
I ripped apart their SEO efforts

I left Twitter 6 months ago because I felt they were holding my own content hostage and demanding a ransom from other search engines to access the data.

As a concrete example, when Matt Cutts first mentioned evaporating PageRank and the “reset vector”, there was a fair amount of chatter on Twitter, including some tweets from me. You won’t find those conversations now.

So why would I give them more data directly?

They have no respect over your content, my content… anyone’s content


Use Topsy – they have much more accurate tracking of tweets, handle 301 redirects pretty well and even give you some really useful historical Twitter data for a whole domain.
Here is Andy Beard on Topsy – that is a vote.

I am not saying everything is perfect with Topsy – they are not doing silly stuff with Robots.txt but their nofollow of links is inconsistent and might benefit from a threshold based on authority and an understanding of first link priority.

There is only one significant benefit of the new official retweet button, you can now define retweet text, which is something I wanted Tweetmeme to include a year ago.
Combined with custom URLs in theory you can use that for split testing – in practice I don’t think it is going to work as there is no way to define a canonical URL, plus a URL that gets tweeted with a URL decorated with tracking parameters.

Thus the new Tweet button sucks for marketers even 1 year on.

More on Mediagazer… this seems to have rolled off of Techmeme in less than 24 hours.


Just adding a test new button that has been modified manually

As text link: You should Tweet This Post

Test without defining the title

As text link: You should Tweet This Post

With Nothing Defined

As text link: You should Tweet This Post

A Shortcode Test

Liked this post? Follow this blog to get more. Follow


  1. says

    1. You can add nofollow to their anchor tags if you want. It doesn’t affect the results.

    2. Their search is improving, slowly. I get results for my domains, sometimes.

    3. Their count only goes back to July. They mention this in the developer documentation.

    4. No argument.

    5. I unfollow people who don’t retweet the right way. I block people who RT me the wrong way. Seriously, just stop doing it wrong. If you want to add commentary to a tweet, then use a reply with something in front of the person’s name (so your followers will see it too). A reply adds metadata that links back to the original tweet, thus giving props and link-love to that person. RT’ing the wrong way is content theft, plain and simple. Doing it wrong also makes it difficult to follow conversations. If it was up to me, I’d have them block Twitter clients that support the wrong way of doing it.

    6. Actually, their “data” attributes are standard HTML 5 and will validate just fine. Switch your doctype and it works. See http://ejohn.org/blog/html-5-data-attributes/

    7. If you examine the javascript, it doesn’t give any data back to twitter. It’s just a simple conversion from the anchor link to the button iframe.

    • says

      1. Adding nofollow still in theory leaks juice, at least based upon Google’s (Matt Cutts) explanation last year.
      Other buttons are pure javascript that is normally blocked with Robots.txt – in general, widgets that are universally blocked with obots.txt Google doesn’t seem to accunulate PageRank on the URL, even though technically they could.

      2. Their search isn’t improving

      3. Whether they mention it in the developer docs or not, they don’t get the current count right even for current tweets using their own shortener, and historical tweets is vitally important.

      5. Link Love? On Twitter? They block most links other than spammy lists with nofollow – lists are games for porn pills and casino SERPs like crazy, whereas conversations with people I respect get nofollowed links. I also noticed Twitter are giving live links to Google maps on every GEO-local tweet – the maps searches are blocked with robots.txt
      The official RT method doesn’t give more meta data – as an example this post was tweeted by @andybeal from Marketing Pilgrim – if I was publishing my content to Twitter (I am not), then he might have retweeted me – if people had seen his retweet and used the official method, I wouldn’t get the data that his tweet had influenced 10+ others to retweet the story. Twitter might have that information, I don’t.
      One of the points I raised in the past on retweets, and specifically in relationship to buttons is that as a user clicking retweeting buttons, I still want to be able to measure my own influence in retweeting that link, and to use a URL shortener that is my own preference to manage that data.

      6. I wasn’t aware it was valid HTML5 but looking at the code being used on Techcrunch there also seems to be some unspecified “features” that make some of the decisions slightly more reasonable, though it still shouldn’t be a link.

      <a href="http://twitter.com/share" class="twitter-share-button" data-url="http://tcrn.ch/bqbXaK" data-counturl="http://techcrunch.com/2010/08/13/five-viral-rules-video/" data-text="Five Rules For How To Make Things Go Viral&nbsp;(TCTV)" data-count="horizontal" data-via="techcrunch" data-related="erickschonfeld:Author of the post">Tweet</a>

      I spy “data-counturl” which does make a significant difference.
      7. The javascript returned doesn’t have to do anything – the subdomain is .twitter.com, they know who I am when I am logged in or have cookies, they have logs and for any company not to use it for data in some way would be silly. There may be a way to host the file yourself… though then we hit fun licensing issues.

      There are a few other issues I need to check out, but I keep seeing errors from the twitter share page

      • says

        The attribute of data-anything is legitimate HTML5. The data-* is meant to be freeform.

        The reason for it being a link is to make it work even if Javascript is disabled.

    • says

      I should add I am really not a validation Nazi

      I have a project which uses a data element with a span multiple times on a page, and there isn’t a timescale for when that will be officially recognised. I was assuming that it was the same with links which was a mistake.
      The same project is also an accessibility minefield.

      One viable method would be to use a dynamic image link which only shows if javascript is off. Then the link would have some point in being there, and the landing page could be generated dynamically to offer both information about buttons, and some search results based on the referrer.

      Twitter are not the first major site to have a badge of some kind rewriting a static link, Technorati did it… but they linked through to the profile of the blog on Technorati. That is user and search friendly.

  2. says

    Hi Andy,

    I’m one of the developers of the Tweet Button and I thought I’d try and shed some light on some of these issues.

    1. Nofollow

    As Otto pointed out, you can do what you like with the link. As long as it’s got a class name twitter-share-button it works. The goodies page is designed to provide a very simple implementation but if you want you can adapt everything to suit your needs. We’ve built a huge amount of flexibility into the implementation with multiple entry points depending on your needs.

    I find it odd that you say that the destination of the link isn’t related to the button. It points to the share follow and allows you to share the referring page. We designed this carefully so it will degrade gracefully for users who, for any reason, do not have javascript enabled. All this aside, you do not have to use the widgets.js / anchor tag method at all. The documentation on dev.twitter.com shows a way of embedding the button directly.

    2 and 3. History and Count Accuracy

    As Otto pointed out, we started counting in July but we’re investigating the possibility of back filling. I agree that historical data would be great to have.

    4. Longevity

    The actual link in the Tweet Button to the search page is http://twitter.com/search?q=term which, on the twitter.com, redirects to the URL you see. As the site changes over time /search will remain a permalink for search. But yes, our search feature is very temporal by design. Twitter search is there to provide real-time, relevant tweets about a subject rather than a historical record. Twitter search on Google or Bing is better suited for archival search. We provide our content to third party services so they can build different types of services like this.

    5. Broken Retweets

    Enforcing any style of retweeting was not our intention with the link expansion feature. The feature is designed to give users visibility in to what they are actually clicking on. We are aware of the copy and paste issue and are working on ways to address it.

    6. Validation

    As Otto pointed out we are following a standard and have taken great care to make use of all of the appropriate emerging standards rather than go with anything proprietary.

    7. Data

    Yes, we can obviously see referrer data that we could use to monitor the take up and usage of the button. However, sites that do not want to give us visibility of this data are free to create their own Tweet Button that simply uses our URL sharing endpoint. You will find integrations like this out in the wild already. Our documentation explains how you can do this. The Tweet Button is all about helping people to share meaningful and interesting content easily.

    • says

      Hi Dan

      Thanks for taking the time to comment – I have a followup blog post which will address some of the issues.

      I have been exploring some of the flexibility, so for instance you could easily have had a button that validates and has significantly reduced SEO risk. To be quite honest you wouldn’t get anywhere near as much negativity if the main Twitter site the SEO implementation actually made sense.

      On count accuracy Tweetmeme has improved a fair amount today, as if they have sudden access to a cleaner stream of data. It still don’t quite keep up with Topsy but it is close. The Twitter count is way off still for Shaun’s earlier post.

      One of your collegues, @rsarver was suggesting on Twitter that a history of reactions is a different product. I can’t understand that line of thought. The tweet count is based upon the historical tweets – if you don’t have the historical tweets, the Tweet count is wrong.
      Twitter’s ability for someone to click a button to look at historical tweets is under question. It is half the function of the Tweetmeme & Topsy button – the other half is the count based upon that historical data.
      If I was running just the new buttons, and in 3 weeks time someone came to this post, will it show 50 tweets or just the one or 2 that might happen later on? Even if you store the count, when someone clicks the button, will they see just the most recent 7 days?

      For me, the actual process of tweeting is secondary to the social proof of the count, and the history. That is what differentiates dynamic tweet buttons from historical static ones. That is the special sause that made tweetmeme popular.

      The validation problems are fairly easily solved (plus I am not a validation nazi) – some of your partners actually have valid solutions – the issue is really about your default buttons not what is possible with the API.

  3. says

    This is an excellent article and I had not even considered the lack of nofollow in the Tweet button itself. I had been using Tweetmeme previously, and am using the Twitter button as part of an experiment, but may switch back to Topsy. Lack of analytics for t.co links right now is pretty compelling reason to go back to using something relying on Bit.Ly where you can see analytics just by adding a + at the end of the URL.

  4. says

    Regarding missing tweets in old searches, Twitter admits searches are limited to recent times. But various other search tools should be able to find all the old tweets. There was even a tweet recently from @Twitter listing a dozen or so such search tools. I highly doubt anything was actually censored. And if it were, I’d sure like to know about it.

  5. says

    Awesome article, I personally like the tweetmeme retweet buttons. I’ve gotten some articles to go viral and that brought in some decent traffic and good link building.