I want to preface this with a note that I do not know how any of the current Twitter analytics providers and reporting services operate. All of what I’m about to say is conjecture, based on what I know about web analytics. I make no claim about any specific service. This description only aims to convey how I might do what they do and investigate some of the hypothetical emergent behaviors.
This is a thought experiment, so let’s first consider the system mechanics. I post a link to my Twitter feed with some short summary of the content contained therein. Some subset of my followers (along with a handful of folks who are not followers) will see the link and click on it. I can easily track the time of my post. Some URL shortening services offer the ability to track all the URLs you’ve shortened and provide analytics data on those who visited your links. bit.ly, as an example, includes a real-time report of number of clicks for each shortened URL. Using these two metrics, I should be able to draw some conclusion about links I share on Twitter, as long as I use a shortening service that offers analytics and make sure to only share links I’ve pre-shortened. Kind of a hassle, but worth it if you’re really serious about your return on investment (ROI) in social media.
There are services to correlate my followers’ activity in an attempt to determine the best time of day (for me, given my current followers) to tweet to reach a maximum number of people. Presumably, they’re hunting through my followers’ feeds attempting to define the periods of time when most of my followers are actively using Twitter, and thus most likely to see what I post. This strikes me as dubious. Sure, there is likely to be a peak period when they are most likely to be using Twitter. It’s even plausible to infer from their mention history whether they are engaging in active dialog with me most often. What strikes me as implausible is that any conclusion can be drawn about link-following habits simply from engagement metrics. It’s more a measure of when folks are likely to reply.
The way I define success in this instance is by comparing the number of clicks I see on a shortened URL against the number of Twitter followers I have. Clearly, it’s possible for this ratio to be greater than one, such as if every follower clicks the link and some random user (outside the immediate network) clicks the link. This can be exaggerated by re-tweets, quotes, etc. There truly is no upper bound on this value, but higher numbers are better. The goal is to reach as many people as possible with a single tweet.
I believe there is a significant time component involved also. The amount of time that passes between posting and clicking seems to be fairly short in most cases. In fact, it strikes me that there must be an average time-to-live for Twitter messages of less than 24hrs. I’m using time-to-live as a value describing the average elapsed time between posting and the time representing the 90th percentile of clicks. In other words, the start time is posting time and the end time is the time by which point 90% of the clicks that will ever occur should have occurred, and chances are that falls within the first 24hrs after posting. I’m pretty sure this is true, but I have no hard evidence to support this claim.
This is the sort of social science I can get into. If only I had the time or patience to focus on it more, I might actually collect some hard evidence and have some more serious rigorous approach. Then, maybe you might believe I know what the fuck I’m talking about. Yeah, I wouldn’t trust me, either. 😉