The use of hashtags on Twitter has expanded significantly and become much more common since the tool was first introduced – by users – as a way to get around what was then a lack of native search functionality on Twitter. But while more and more people are using them and they pop up in magazine ads, TV shows and other media as a way to encourage online conversation their are many examples of usage that seems to be more an effort to be part of a trend than as part of a strategic plan.

Having slowly adjusted my own thinking on the use of hashtags over the years I’ve begun to see more ways they can be useful for specific purposes, though I continue to maintain their ugly and interrupt the overall user experience. Here are a list of points to consider and keep in mind while pondering how to use hashtags as part of an overall publishing program:

  1. Don’t do it just for the sake of doing it: Not everything needs to be hashtagged, plain and simple. There are times and places to do it but they are not necessary just for search anymore and overuse only adds to the creation of an ugly stream of updates. Think of them like Pink Floyd records: There weren’t a ton of them but each one meant something. This ideally applies to publishing as a whole and doubly so for a tool like hashtags. 
  2. Make sure it’s contextual: This is pretty simple. Does the hashtag that’s being considered mean something within the context of the program?
  3. Makes sure it’s easily understood: Similar to the one above, people shouldn’t be scratching their heads over what’s going on here. There should be no Python-esque “What’s all this then?” moments in the audience when someone comes across a hashtag that’s being used. 
  4. Define the goal: Some of the best uses I’ve come across are when they’re part of an event or used to collect contest entries. The audience needs to know in short order why that hashtag is being used and to what purpose. Sometimes that can be simply to aggregate conversation around a single topic – the conference/event example is a good one here – and sometimes it can be to elicit responses. 
  5. Listen to the conversation: This is just a microcosm of what should be an overall listening program but it bears repeating that a conversation that’s not monitored and listened to doesn’t fully qualify as a conversation. 
  6. Decide on how the conversation is used: How is the feedback that’s asked for being collected? How are the entries to a contest being collected? How is the conversation being aggregated or repurposed? Most importantly: Is the audience aware of the ultimate goal (in as much as it makes sense for them to be so) of the conversation they’re being asked to participate in? 
  7. Define the metrics for success: As with most things the definition of success needs to be agreed upon beforehand. This may not include actual numbers, particularly if a tactic is just being tested out, but some sort of level needs to be set at which point everyone agrees “Yep, that worked.” 
  8. 8) Consider the actions if things go pear-shaped: The last three or four months have seen a number of instances where a company will try out what they think is a fun hashtag tactic only to see things get weird quickly. I mean things can really escalate fast. So while the odds may be low that some disgruntled members of the audience will hijack the conversation and create exactly the wrong actions there still needs to be a plan should that happen. This plays directly into the above point on deciding what is done with the generated conversation. If hashtagged updates are being fed onto another webpage or into ads or something there better be a level of moderation/vetting before that happens.

While all these points are absolutely good things to keep in mind and work out ahead of time I will say, in an effort to contradict everything I’ve just written, that there’s also something to be said for trying an idea out a spur of the moment manner and seeing what sticks. Like most aspects of status network publishing it’s easy to iterate quickly, testing out an idea and then deciding if there was enough traction to continue using it. But that doesn’t mean you get to skip the steps outlined above, just that the process behind them is somewhat compressed.