Automated Sentiment Analysis via the social web is a relatively new field. The idea is to analyze status updates, tweets and blog posts on topics and assign a positive, negative or neutral view of them. It relies on analyzing content through Natural Language Processing (NLP). Maria Ogneva at Mashable recently wrote an excellent article on the topic and outlining the technology, but as for usefulness for most businesses we disagree.

Our view is to beware and not get star struck by the idea of Automated Sentiment Analysis from any vendor trying to sell you on it. This is what you need to look know before paying for any Automated Sentiment Analysis system:

1- Are you a big enough brand, with a unique name? Blue chip companies get people talking. If you are a small business chances are not a lot of people are talking about you. A simple search at something like Scoopler.com can let you know. Hopefully as well you have a unique name. Good luck if the name of your brand or company are common words. A bakery called “FreshBread” will have a lot of trouble finding people only talking about their business.

2- Are you engaged in Social Media?
If you are tracking conversations about your brand, it helps if you’re engaged and adding your own voice. For example, if you don’t Tweet how can someone re-tweet and add their opinion?

3- Are you products something easily identifiable?
If you want to track a recent product you have release make sure it has a name you can track. Something unique. Common names, like Bread or Cupcake might be hard to track, unless you have a unique company name.

4- Who are your customers and what is your industry?
If your customers are older, and don’t even have a computer how do you expect them to be online talking about you? Or examine if are you in an industry that even uses social media. Do your competitors use it? I know it’s hard to believe but some industries, don’t engage in social media. Therefore, it very likely no one will be talking about them and Sentiment Analysis would be a waste of time and money.

5- Where are you customers located? Will they be speaking English?
Sentiment Analysis has issues with sarcasm, jokes, cultural factors, linguistic nuances and more. Basically, just there are issues with automated translation, automated sentiment suffers from the same. The only way automated sentiment analysis would work are huge complex algorithms, that have the permutations in the native language and how sentences are constructed. We are far from here.

If you are able to answer these questions with positive answers and still intent on exploring Automated Sentiment Analysis, the one vendor that has been most promising in this field has been Radian6. We believe this tool only makes sense for companies with large brands with a highly engaged audience.

In a nutshell, Automated Sentiment Analysis is a great idea, and can be helpful, but vendors still have a lot of work to do in getting it right.