Rogue Tweets: Where They Come From; How to Stop Them

What do the American Red Cross, the US Secret Service and the Chrysler Corporation all have in common? They have all been embarrassed in recent months by rogue tweets, those unfortunate errors that often go viral and frequently cost someone their job.

Rogue tweets are caused by a variety of reasons. Sometimes it is an accident caused by software that automatically shortens content generated for other purposes, as in the time the BBC tweeted "HSBC, Europe's biggest bank, reports a 24% drop in annual profits for 2009 after writing down the value of ass." This has to be one of the funniest truncation errors ever.

Another classic accident is to confuse the tweet and direct message windows, the cause of many errors like the time when author Rebecca Skloot tweeted "My cell phone is xxx xxx xxxx, what's yours". Poor interface design is at fault here.

Sometimes it is the fault of someone scheduling a tweet with placeholder text that is never replaced before publication. Organizations that schedule PR activities long in advance are most vulnerable to this, like the time NASA tweeted "NASA to ship fuel tank for the last planned shuttle flight. PASTE INTRO SENTENCE HERE." To be fair this used to happen often in press releases so there is nothing new here.

Automatically sending the same content to multiple social media platforms causes problems, too, as in the case where a consultant tweeted "Are you on Twitter? Here is my page...". Yes, Robin, here on Twitter we are all on Twitter.

But by far the most common cause of embarrassing tweets is when people confuse their work and private accounts. This explains three of the best known episodes in recent months.

First was the American Red Cross that accidentally went viral with the inspired tweet "Ryan found two more 4 bottle packs of Dogfish Head's Midas Touch beer... When we drink we do it right. #gettingslizzered". Not long after that came the Chrysler Corporation that tweeted "I find it ironic that Detroit is known as the #motorcity and no one here knows how to f___ing drive." Finally the US Secret Service tweeted "Had to monitor Fox for a story. Can't Deal. With. The. Blathering.

In all of these cases what has happened is that the person who is supposed to be tweeting for a company account has believed that they are tweeting on their personal account. This can happen even when you have separate clients for the two accounts, or even when you are using a desktop for the work account and a mobile device for your own. But people who work in social media agencies or for in-house social media departments are normally using clients or web services like Tweetdeck or Hootsuite where each account is just one column, and you always type the text into the same window, selecting the account by clicking on small buttons. If someone adds a column for a personal account the risk of cross posting is extremely high.

You can reduce this risk in several ways. First of all you need to put in place and apply very strict policies to make sure that people who are working on your account are not adding a personal account column to the work client. You should also restrict any tweeting to their personal accounts at the same time even on personal mobiles, perhaps by allowing it only away from the workstation or not at all. The kind of people who are best suited to this kind of work are almost always very active on their private accounts so this might seem harsh, but there are plenty of other jobs where you don't tweet as you work -- airline pilots, for example.

In addition it is prudent to use whatever checks that your software tools can provide. Hootsuite, for example, has an enterprise edition that allows you to limit the permissions of employees, so that juniors can write and schedule tweets but they are not published until someone with more authority has validated them.

I still think, though, that the makers of social media clients could add even more features to catch rogue tweets. Perhaps adding a dictionary of words like "beer" and expletives so that if someone tries to tweet them it flags at least a warning like "You just asked to tweet about BEER to the account VATICAN. Are you sure you meant that?"