Yesterday I stopped by Twitter HQ to see Dick Costolo. Dick recently moved to Marin (my home turf) and took Twitter’s COO job. I’d say that taking such a job means Dick’s hair is constantly on fire, but if you know Dick, you know that’s really not an issue. (He’s level headed, he’s a pro, and, well….let’s just say he doesn’t wear his hair long).
Among other things (FM has partnered with Twitter in the past, and will continue to do so), we discussed how Twitter might crack the code around explaining its user base, how those users engage with the service, and how the service is growing – especially given the recently hot (and to my mind not well understood) topic of *if* it’s growing. Dick assured me it is – echoing a recent tweet from founder Evan Williams.
Much has been written around the topic of Twitter’s growth, but the shorthand is this: You can’t rely on Comscore or web-based measurement services like Compete or Quantcast, because they do not measure the entire Twitter ecosystem, which is distributed in nature. For example, these services do not measure use of Twitter’s API, which accounts for more than half of the service’s traffic (through apps like TweetDeck, Twitteriffic, Exectweets or Stocktwits, for example). They also don’t measure mobile usage, and some don’t measure international traffic, which Costolo said in some countries is growing “straight up” – quite like it did in the US early last year.
Regardless, these services do show a flattening of traffic to the US domain, which if not explained, will continue to cause consternation and questions around whether the Twitter ecosystem is indeed continuing to flourish. And when the service begins working directly with marketers, those questions will need to be addressed. Not to mention the issue of inactive accounts – folks who join but don’t understand how to extract value from the service – witness Radiohead, as one example. A lot of folks come to the service, tap the microphone, ask “is this thing on?”, and then leave. Lists and a revamp of “suggested users” was the start of the company’s fix to this issue, and Costolo told me he has a team focused on next steps.
Dick also mentioned that there are a lot of folks who use Twitter to consume information, rather than broadcast it. Those folks are valuable audience members, but it’s hard to prove consumption without a metric to validate it (IE, number of times a user pulls a Twitter feed or visits his/her page).
While I don’t have any news to report on whether Twitter will be releasing its own stats, Dick shook his head emphatically when I asked him if the recent Royal Pingdom post about growth in overall tweets was directionally correct.
According to that post:
According to our research, Twitter is as of December processing more than one billion tweets per month. January passed 1.2 billion, averaging almost 40 million tweets per day. This is significantly more than Twitter was processing just a few months ago.
While understanding “tweets per day” is a fine metric, it’s not very deep. For example, perhaps a very small number of folks are creating most of the activity (a problem that Digg has had, though it does get a lot of “hummingbirds” – folks who come and consume one quick page, then leave). A more valuable metric would be “active users”, a standard that both Facebook and MySpace have promoted over the years (as one might expect, MySpace hasn’t really promoted that particular metric much lately).
Dick agreed, but doesn’t have anything to announce on that measure, yet. We did discuss how having such a metric will be important for the company once it rolls out monetization. (That Twitter intends to work directly with marketers is certainly no secret.)
Other valuable measures we discussed were engagement and resonance – or how a meme travels through the Twitter ecosystem, through influencers, retweets, and co-incidence (IE, lots of folks tweeting the same idea/URL/meme at the same time). Brands are particularly interested in understanding this ecosystem, for crisis management, identification of brand influencers, as well as “listening” (and responding) to the conversation on topics related to their products and services. Many listening/engagement services exist (Radian 6 , Converseon, etc.) but Twitter itself could be doing a lot more to surface useful data that services like these can leverage. (And, btw, if you want to create TweetSense, having these measures will be crucial).
I expect more from Twitter soon on these fronts. If you pay attention to Twitter’s blog, you’ll have noticed a recent post titled “Super Data.” Study this and you see an EKG of sorts for ads during the Super Bowl. At one point, 19% of all tweets were about Doritos. That’s co-incidence for ya.
The post noted: “There is real value in being able to measure the reach and influence of those topics in real time, and we in the analytics team are looking forward to a lot more where this came from.”
If that’s not a hint that more is on the horizon, I don’t know what is. What kind of measurement and metrics would you like to see from Twitter?
Meanwhile, here’s your Friday linkage:
Why Brands are Becoming Media (Mashable/BrianSolis) Yes yes yes. I have been on about brands = publishers for a very long time. Yes.
Should We Clone Neanderthals? (Achaeology.com) Look, sometimes I need to toss one of these in here to see if you are paying attention. So should we?
Survey: 1 in 5 marketers to shift 30% of traditional marketing budgets to social media in 2010 (Socialmedia.org) Again, I just like stories that confirm what we all are eager to see happen.
Lookin’ for Love in All the Wrong Places (AdAge) Brands need to get more emotional in their decisions. I call that being human!
Spatial Search: The Next Frontier (Bing) Cool new stuff from Bing.
comScore Releases January 2010 U.S. Search Engine Rankings (Comscore) And…Bing is gaining share again.
Yahoo Display Strategy Turns To OPA Ad Formats In Effort To Drive Premium Prices (PaidContent) Well, yes, FM adopted some of the OPA standards, and it’s nice to see Yahoo follow suit, but we took those units and made them better by making them social.
The Numbers: Super Bowl Ads, Social Media (MarketingProfs)
I was somewhat surprised to see Twitter’s blog post about Super Bowl stats, since they’ve traditionally been so reserved about publishing any kind of actual numbers. It was an interesting insight into how trends on Twitter actually work however, so I hope they do more posts like that.
As a marketer, the stats that are most interesting to me are those about engagement. Part of this is to combat users that just inflate their numbers: I don’t want to see how many people are ‘following’ you; I want to see how many people are actually engaged with what you are saying.
Lists have made it a lot easier to see who has inflated numbers, and who has followers that are actually paying attention. I think additional information like the % of a user’s tweets that get retweeted and the % of a user’s followers that have @replied that user would also go a long way towards figuring out if a user is actually influential or not.
I don’t care how Twitter tracks for themselves, or for PR purposes.
I want to track my Twitter pages as landing pages.
I want to define goals such as new followers and where they came from.
I want to be able to add more text so I can comply with UK law or have more hyperlinks even if they insist on nofollow – with descriptive anchor text
I realise they need to somehow have support in other platforms with their character limits, but I am sure lots of mobile users are now using dedicated apps rather than SMS, and there isn’t a need to be able to retreive all info to SMS
This is a good post and I like the comments, too. Since you wonder what sorts of analysis could be done with tweets (e.g, those related to Super Bowl), check out TipTop’s analysis of Super Bowl ads at http://ftt.nu/superbowl While there, you can also check out their other Specials that are also based on very sophisticated analysis of millions of tweets.
A personal metric of mine that I’ve been tracking is tweets per follower. Over the past month, it has gone from around 8.9 to over 10. Assuming I tweet the same amount per day, this shows that the followers I’ve gained has either leveled off, or, for whatever reason, I’ve become more annoying. or more likely, less influential.
Only time will decide which is more correct.
We definitely should!
Before I read much on social network marketing I was thinking the quantity of link data in a tweet was very important to Google’s Page Rank for example. Tweets are FULL of link data. I’ve been amazed how much info you can get into a single Tweet:
http://twitter.com/m_holloway/status/9075897798
One tag was not exploited in that tweet was the ‘in resonse to’ element.
Upon further investigation I learned the qualities of data (like how and why tweets spread) are just as, or more important.
But I’d still like to see a comparison of numbers of links per hour or day in different social spaces. Is there a resource?
Michael Holloway
I have used external apps that show users tweet stats in depth. For example, it is interesting to know which followers tweet links 100% of the time, or how seldom someone retweets or @replys. These metrics provide powerful insight to brands who look to build a targeted and engaged group of followers.
If twitter metrics also included semantics you could easily get a feel for a twitter user based on the tag cloud of their tweets. This person mostly tweets about ‘Sports’. I am looking forward to leveraging those types of metrics as part of our deliverables for clients that hire us to do Social Media Marketing.