If you happen to live in America, chances are you were aware of the big Super Bowl game on Sunday. If you were on Twitter during the Super Bowl, you may have been aware of Brand Bowl (#brandbowl) – a movement fueled by social media enthusiasts who tweet their opinions on the TV ads that air during the Super Bowl. (Brand Bowl is a partnership among Mullen, Radian6 and Boston.com) What was derived from all that tweeting was the best and worst Super Bowl ads of 2011 according to the #brandbowl participants. And most of that was possible thanks to a specific type of analysis – sentiment analysis – sometimes called opinion mining.
Having spent over 3 years at Lexalytics, a software company that offers sentiment software, and by currently assisting in the Sentiment Analysis Symposium in NYC, I’m pretty familiar with sentiment analysis. And with the birth and explosion of social media monitoring tools and software there has been a high interest in text extraction and sentiment as it pertains to marketing.
I thought it would be useful to outline some of the factors in a social media monitoring solution that you should consider if you are including social media as part of your marketing plans. You’ll want to keep some of these in mind as you determine how you will measure the effectiveness of your social campaigns.
- Content: You need to think in sources and volume. Will I be measuring just online social media or do I need to include other data sources? How much content do I expect to analyze? A few hundred documents, tweets, posts or millions of them?
- Accuracy: Do I need to be 99% accurate with my results or 80% accurate? This varies widely depending on what you are using the results for – is this to measure customer satisfaction? Market/PR messaging? Crisis management (see Kenneth Cole recently)? Brand awareness? Product Launch? In most cases you can automate some sort of the process by eliminating the bulk of the “neutral” data and focus on the extremes.
- Depth: Are you analyzing content at a document level (i.e. This post was good for our company.) or at an entity level (i.e. This is bad for our CEO but good for the company).
- Cost: Price always comes into play. Are you looking for a light service or a robust, enterprise solution?
- Customization: Will you need to analyze unique data that isn’t mainstream? If your business is widely familiar to the masses, chances are a monitoring solution can handle the content. But if your company had created a unique product with a very targeted audience (i.e. pharmaceutical, biotech, government) then there may need to be some customization in analyzing your results. But don’t be dismayed – many solutions can help you get a handle on the social media content.
The nice thing about sentiment analysis is that there are several ways to apply it – there are automated solutions and there are manual solutions. You may find a mix of the two works best for your needs. But one thing is for certain, your customers and prospects have a lot of opinions about your products and services and they like to share them online, so it’s always a good idea to be listening.
And by the way, my favorite Super Bowl ad was a tie – Darth Vador for VW and Reply All for Bridgestone. What were yours?