David Bradley, otherwise known as @sciencebase, put together this adorable heuristic flowchart modeling his decision on whether to follow someone back when they follow him. Says Bradley:
How do you decide whether to follow someone who has followed you? There are some basic filters you can use, like not following back obvious spammers and scammers and generally not following people with protected tweets unless you know them already.
I’ve created a flowchart to help you decide whether to follow someone who followed you on Twitter (click the image to get a fullsize view).
He then goes on disclaim, presumably for the benefit of readers who think they might not make it through this gauntlet:
Incidentally, I am not quite so strict as this flowchart implies so please do follow me as @sciencebase
Humorous. He’s actually probably quite liberal in who he follows back; he’s basically just checking for “genuineness” and “humanity”.
It turns out that this is a very good heuristic in general for an individual or a highly socially engaged business to follow, including the “maven, guru, entrepreneur, expert” and “lack of irony” accusations, when they want to vet new followers.
It also turns out that it’s really pretty easy to program a computer to determine which Twitter users are bots (automated) and which are human to 95% or better accuracy. (Why most bots out there, many of them attached to services people are actually paying for, don’t make this distinction — and why it’s totally stupid and counterproductive that they don’t — will be the subject of a later post. –ed)
Much more difficult, as you might imagine, is determining “boringness” and making judgement calls on the “lameness” of an avatar. Certainly the lack of an avatar can generally be deemed “lame”, but where do you go from there? You start getting into the realms of computerized image recognition and computational linguistics, that’s where, and then it gets complicated and processing-intensive. Not that we’re against that. But for the purposes of this discussion, we’re going to stick with the low-hanging fruit.
Twitter provides quite a bit of low-hanging fruit to pick and process into some delicious heuristic jam. Through the Web and through Twitter’s API, anyone can be provided with quite a few more indicators that help clue us in to our subject’s purpose in being on Twitter and their attitude toward engagement overall.
Tweets themselves qualify users in their personalities. Here are some of the features of Tweets we use to model individuals on Twitter and evaluate their potential for interaction in your market:
- @mention (reference to another individual on Twitter)
- @reply (= @mention in the first part of the Tweet not shown to others who aren’t following the @mention’d party)
- ~@reply (= @reply deliberately exposed to people not explicitly following the @mention’d party)
- #hashtag (display of interest in an ongoing subject… OR an ironic gesture)
- URL (full) (plus, whether the URL is accompanied by some other text or not)
- URL (shortened — and type of shortener is important — see Brian Zarrella’s brilliant Science of Retweets report)
- [blahblahblah] (contains only a statement, none of the above features)
- [...] (nothing. lack of meaningful information is a significant feature)
Going to the trouble of marking off Twitter users in a more rich fashion goes a long way in producing meaningful quantitative data to inform your niche marketing efforts. If your Twitter outreach results in 100 new followers who have mentioned “wine” in their bios, but who are 30% bots and %50 “social media gurus”, those 20 left over may well be worth your time — but how will you figure out how much time it will take to furrow out the false matches to the real customers?
AND, once you’ve gotten down to those 20, how do you feel about a conversion rate of, say, 5% bringing your total ROI on those 100 new followers to a mere 1 individual, if that?
If there were a tool to eliminate the need to spend the time and effort to investigate 100 leads just to acquire 1, would you try it?
Loading...