Is Social Media Worthy of Text Analytics?
Coke Says No Boost from Social Buzz – Should You Care?
Guess what, “Coca-Cola recently learned that Social Buzz doesn’t correlate with it’s short term sales”.
Here’s how it works for those of you who don’t understand marketing, social media or analytics.
Social Media is a tactic – a tool. If you have a great advertising campaign wherever it is on TV and Social Media, just on TV, just on Social Media, whatever… then someone is probably likely to mention it on Twitter or blogs no matter how inconsequential to life your product is.
Conversely, if your advertising campaign sucks they are not likely to mention it much. Well unless it really sucks then a few might say that.
But so what, when it comes to predicting sales, monitoring tweets is at best looking for a trailing indicator – not the cause. Surely your sales numbers are a better indicator of your advertising success?
Social media certainly can and should play a critical role in most marketing campaigns today.
However, please, please realize that “Social Media Listening”, constitutes listening to mainly spam from the 8% of our population who tweets and blogs (while technically far more representative no one can analyze Facebook pages as they remain private/accessible only to Facebook).
Researchers, let’s keep things in perspective shall we. Way before we start discussing which approach to text analytics is most powerful we have to first decide what data is worthy of analysis. 140 character tweets from 8% of the population mainly trying to sell their “expertise” is just about the poorest form of data out there. Garbage In – Garbage Out!
The fact that Twitter even scores as many mentions as it does for products like “Coca-Cola”, which most regular consumers would be unlikely to ever think about any given week, is that there are so many want to be social media marketing guru’s on Twitter and blogs trying to analyze others marketing campaigns – further proving what a peculiar sample blogs and twitter is.
If you actually want to predict something like sales, you will need to first have a long serious think about what data, whether unstructured or structured, makes most sense in your model!