Sysomos today released a study of over a billion tweets that detailed the following:
71% of twitter messages fail to produce a reaction (reply or retweet)
85% of Twitter "conversations" are only one level deep
92% of retweets happen within an hour of the initial tweet - in other words, if it isn't retweeted within the first hour, a tweet will likely go unretweeted (I feel a burning shame at even writing this non-word word).
This is a good piece of work, and I've seen a lot of people, appropriately enough, retweet this study and its coverage on sites like Mashable, who provided the helpful headline "Most Tweets Produce Zero Replies Or Retweets."
There are a number of ways to think about this study. One way, as many have already done, is to produce comment after comment about how this proves Twitter falls on a spectrum between "low engagement medium" and "utterly useless."
I see a different message.
I know I write a lot in this space about how mining clickstream data alone will never get you the right answers - and that combining online AND offline inquiries is the only way to gain real insight into the people who are writing and reading content on sites like Twitter. Yes, I know I work for a company that gets hired to ask these kinds of questions, and you might be tempted to dismiss my thoughts as promoting survey research for selfish reasons. But hear me out.
If you judge the effectiveness of Twitter on the basis of data like the Sysomos study, you are left with one inescapable conclusion - most people don't actively engage on Twitter. This is a demonstrable fact - there's nothing new in that insight. Judging the platform, however, by how few (or many) individuals respond/react to tweets would be like "measuring" the engagement of the New York Times by how many people write letters to the Editor - or worse, judging the circulation of the Times by this measure.
This study gives you some new things to know, and I believe them. Here is what you don't know about Twitter:
How many people had the "opportunity to see" a given tweet?
How many people actually saw a given tweet?
How many people saw a given tweet and could recall its message?
How many people recalled a message, and told a friend offline about it?
How many people saw a message and took some kind of offline action in response?
You can't answer any of these questions by mining for tweets - and yet, I can tell you those answers and many more about a wide variety of traditional media that aren't throwing off 1% of the raw "data" that Twitter is. Radio, TV, and yes, the New York Times all know the answers to these questions, because they are salient to their business.
They are unknowns, if you rely solely on clickstream data - but they are not unknowable unknowns. Yes, I get paid to ask these questions for clients, but I am passionate about asking them. It's time for social media practitioners to start asking better questions. If Twitter is, in fact, a broadcast medium (and Twitter themselves are basically telling you that they are), then asking - and answering - the questions I've posed above is the only way to know if your tweet fell on deaf ears.