On automating Twitter

toolsI recently coded a Tweet bot to experiment with automated or scheduled Tweeting … because, well, in today’s cut-throat on-line social media marketing world I’ll get left behind if I don’t, right? Only joking.

The bot is a simple PHP CLI script which ties together the codebird Twitter API library, an SQLite backend for storing Tweet data and a directory full of images. The script runs as a cronjob in the cloud and works a treat technically speaking.

The purpose of the bot is to promote a blog by regularly tweeting links to blog articles. The tweets contain text including several hashtags, an image and a URL. Twitter requires 23 characters for a URL and 24 characters for an image, so the maximum number of characters available for text and hashtags is 93. The bot tweets a repeating monthly cycle of 31 tweets, at the rate of one tweet per day. Every seven days an advertisement for the web design business that built the blog site is also tweeted.

Although the Tweet bot hasn’t been flying for long, a couple of opinions on the exercise can probably be safely ventured.

First of all, the activity appears worthwhile. Tweets are being seen, retweeted, liked and links are being followed. Blog articles are not disappearing into oblivion the day after publication, but are now being “bumped” back into view once a month per article. This interval seems legit, the content doesn’t get lost but the tweets don’t become spammy either.

Pre-defining a structured set of Tweets is a useful thing. Whether the cycle is weekly, monthly or yearly, structure ensures that:

  • tweets are balanced in terms of content and timing
  • tweets are sequenced in terms of any sort of narrative that is being created
  • key messages are repeated at appropriate intervals

Spontaneity is great, and random manual tweets and retweets can (and should) accompany the automated tweets, but a structured cycle of tweets ensures that all bases are adequately covered and what needs to be tweeted is tweeted.

That said, automation does not mean a Twitter account can simply be ignored, at least not if you’re serious about building any sort of real engagement. When people engage with tweets, and in particular comment on them and ask questions about them, it is important that they receive a timely response from a human rather than a bot. The complexities of human language and things like irony and sarcasm mean that it is very difficult to automate such responses even if you wanted to.

It quickly became apparent that automatically following back any account that follows you is a mistake. Doing so results in a faster growing follower count but a significant proportion of the new followers will be worthless zombie accounts tweeting garbage. So again if you’re serious about building real engagement with real followers it is necessary to manually follow back genuine followers and ignore the zombies.

To summarise, Tweet automation appears to be a useful marketing tool but if you are trying to build a real business, organisation or community 100% automation is not possible. Automation can put tweets “out there” in an organised manner but any subsequent interaction with those tweets is best handled by a real person.

Postscript: the longer I stay in Twitter, the more I am seeing pointless zombie bot accounts with 37k followers or whatever. Do we really want to add this pointless waste of bandwidth, time and electricity?