Nick Halstead is the CEO of popular Tweetmeme, a popular service that aggregates links from Twitter, as well as providing retweet buttons for thousands of sites.
I’ve been talking to Nick about Tweetmeme, and the upcoming Tweetmeme Analytics service, which is due to be unveiled this week…
Can you tell me about Tweetmeme?
What a lot of people don’t realise is that we originally started Tweetmeme in early 2008, as a side project. We spent a weekend just building this idea about aggregating news through links. We were the first service to aggregate news from Twitter, but we left it at this point to concentrate on Fav.or.it, and it was left to languish for a while.
When we saw that Twitter was becoming more popular and attracting lots of traffic in late 2008 we decided to go back to it, and switched the technology we were using for fav.or.it to Tweetmeme, also rebuilding it from scratch.
We now aggregate 1.5m stories every day, fetching all the various links associated with a particular article, providing a title and displaying associated stories. These are then ranked according to our internal algorithms, which includes factors such as the authority of the story source, the users that are linking to it.
We have a team of ten people working on Tweetmeme, based in Reading.
How widely used is the retweet button?
We have had a lot of success with the retweet button,
and it is now used on tens of thousands of sites, helping publishers
and bloggers to promote their content via Twitter. To give an idea of
the scale, we have had 2bn retweets over the last 30 days, and this
number keeps growing by 20% each week.
Not quite 100,000 sites have the button. It is its own entity really, it isn’t branded with Tweetmeme, though people often know that it’s us behind it. We get thousands of searches for ‘Tweetmeme button’ for instance. People know who we are and what the button does.
People strive to get on the front page of Tweetmeme, as it can drive traffic, in the same way that a front page slot on Digg or Reddit can.
What is the business model behind Tweetmeme?
It has never changed, and has always been about channelisation of content. We categorise content at speed, looking at content and putting it into categories. We can get into some very specific niches, and this is something we offer as a service.
We sell bespoke channels to brands and businesses can pick up on relevant discussions on Twitter. These channels can be hosted on our site, or provided as a white label service that can be integrated into clients’ websites.
It is similar to some monitoring services, but Tweetmeme allows for finer controls. We’re not just looking for sentiment and keywords, the service can allow for semantic targeting e.g. a Britney spears channels could be configured to search for news only on relevant celebrity sites.
Can you tell me a bit about the new analytics service?
This is something which will be launched publicly this week, and we have been alpha testing it for the last few weeks.
We have been running Featured Tweets on the site, where stories can be promoted on the main or category pages, and have been making some decent money from it.
It is effective in driving traffic, and the analytics package comes from this, as a way to show in detail how these Featured Tweets can work.
We are focusing on step by step analytics, so if your article had 60 retweets for instance, then we can tell you how it spread, the click data, who retweeted it etc. Its a paid service for bloggers who want to understand how their posts are performing.
How big a problem is spam on Twitter?
Until version two of Tweetmeme, we didn’t have too many tools to filter out spam, and much of it was done manually. Now we have some filtering tools to help us, and you’d be hard-pressed to find any spam on the new version of the site.
We use a reputation-based system, we look at how long people have been on Twitter, the friends to followers ratios, amongst other factors. While some spammers are obvious, for others you need to monitor and spot patterns that indicate things like automated twitter feeds.
It’s a continuous battle to filter this stuff out, though I think that Twitter itself does a decent job of removing spammers.
What percentage of Tweets are spam?
It’s difficult to say exactly, though we probably remove tens of thousand of spam links every day, out of roughly 1.5m links overall. It also depends on how you define spam, it doesn’t just have to be links for things like Viagra; there’s also some really bad selling going on through Twitter which we don’t want on Tweetmeme.
What happened with fav.or.it?
We shut it down as it was taking up too many resources. Tweetmeme needs two racks of servers to deal with the traffic we get, so we decided to concentrate on this.
Fav.or.it had 2 premises; the channelisation of content, and building a platform to gather together all comments from the web in one place. We decided last year that the latter was not possible, though Tweetmeme does provide a version of this.
I think that the core set of technologies behind it were good though, and this is why we were able to get Tweetmeme up and running so quickly.
What’s your view on the Twitter API? Why has it proved to be so popular?
It’s been well documented that the Twitter API is either a stroke of luck or pure genius to go for something so simple that makes it easy to adapt to. Other APIs, like FriendFeed are arguably more well designed, but are too complicated to attract the attention of developers in the way that Twitter’s does.
Do you worry that Twitter itself will replicate the functions of Tweetmeme and compete with you?
We have a close relationship with Twitter, and engage in a continuous dialogue with them. Someone did once ask Biz Stone via Twitter whether they would ever compete with us, and he replied and said he wouldn’t. I almost tweeted ‘phew’ in response!!
It doesn’t guarantee they won’t do this at some future date, but the growth of Tweetmeme and the traffic driven by the retweet buttons has been good for Twitter as well as us.
Are you concerned about copycat Twitter services?
We had this with ReTweet.com, who used our code, though it wasn’t our back-end code. (as documented here.?
This is to be expected. When Digg launched, lots of clone sites emerged. I think unless someone does something more innovative than Tweetmeme, and does it better than we do, then we should be OK.
In the case of ReTweet.com, they copied a five-month old site which we have since updated, so they are still miles behind us.
How many users does Tweetmeme get?
Don’t give out public numbers, but the figures are in the multiple millions.