Most of you know that I know that I serve as Chief Strategist at Converseon, a leading social media listening company, so I don’t claim to be unbiased in my opinions in social media listening–of course I think that Converseon has the best approach, combining human analysis with technology that makes that analysis scale. But I know that not everyone will spend the money to get their social media listening to be as accurate as Converseon can do it. I know that many of you want to use something free, such as Hootsuite. (I see you out there.) And you can use a free tool to do social media listening–it’s not against the law. But I want you to think very carefully about what you are using it for.
You see, there are two very clear kinds of social media listening. One kind really only needs to look at individual posts–a stream of data that a human being watches on a dashboard and picks out what is relevant. If you are monitoring a crisis, or you are picking out possible job applicants, or you are trying to identify sales leads, this can work just fine with free tools, as long as you are willing to pay someone to sit in front of the screen and watch the stream.
Because in a crisis, you don’t need to see every post, and it”s OK if lots of the posts are irrelevant to the situation, as long as someone is watching and picking out what’s important. If a story is very important, enough people will tweet it that you’ll see it rather quickly. If 95% of the stream is irrelevant to your sales team, but they still catch the few sales leads that go by, it can work just fine, even if they also miss some leads. Free tools can be just fine in those situations.
But whenever you are trying to answer questions that require aggregation of data, the free tools become a lot more difficult to use, because you won’t have the right data to aggregate in the first place. For example, if the cell phone company Sprint wants to judge whether their brand mentions turned more positive when they announced their latest service plans, just putting in the word “Sprint” probably won’t get the job done. In addition to finding all sorts of conversation about their company, they are likely to find lots of chatter about high school races, and they don’t care very much how positive it is.
And if the irrelevant data makes up 30% of the stream, you can’t conclude anything. So, you clearly need something beyond keywords to do your aggregation so that you know that you have the right data. Human analysts can do it. Feeding their corrections into machine language technology can scale it.
So, it’s not that you can’t use free listening tools. It’s not even that you can’t use them to try to answer these data aggregation questions. You can, but you have to do an awful lot of data cleanup to make them work. But that isn’t the way I see companies using them. I see them loading up the wrong data in the tool and shooting out some numbers and thinking that they have answered the question.
I understand the allure of free, but how much less do you want to pay to get the wrong answer?