Web Data Science is NOT Social-Listening

The tonality of the title of my post may sound almost angry. Actually, it's that of a slight frustration pertaining to the marketing sector that has become comfortable with the term "Social-Listening". This post is to clear certain doubts related to it.

I am a Data Scientist with an undergraduate background in Computer Science and Engineering. I have been trained on maths, logic, programming and Data.

Since, the last four years I am working in a wonderful sector called Advertising. People from my kinda background are probably thinking I make predictive models and write code for some "Ad-tech" company but that's not the case. I work in the digital slice of traditional advertising companies who make communication and advertising strategy for global clients at global levels. And yes, if ad-tech is a necessity we incorporate it.

At the heart of advertising are the consumers of the brand. We try and do exactly what the consumer wants, needs, thinks, does or says, which we call "insight".

To figure this not so simple puzzle we use Data. There are hundreds of data inputs related to our brand consumers. This could range from client purchase history to social media user-generated content. Having dabbled with various of these inputs, today I specialize on unstructured web data indicative of our consumers.

The unstructured data comes in various forms -

Conversations on consumer Forums 
Tweets and Blogs
Images and posts on Social Networks
Comments in youtube videos, e-commerce portals
Corporate and consumer website content
.....

The list can be long and it is the work of the thriving unstructured data industry to keep adding to it.

Coming back to the lack of understanding that persists in the marketing sector is that anything to do with unstructured data is often termed as Social-Listening. With which comes the mindset about the work process. Though, anything to do with CRM, Media or Structured Data is good old Data Science or Analytics.

There is a big difference between Social-Listening and unstructured-web Data Science. In a nutshell, Social Listening is performed at the stage after Data Science, when the humongous unstructured data is "almost" cleaned, organized, analyzed and displayed on a sexy interface to the analyst user. Thereafter, a lot of intuition and rigorous manual work is required to make sense of that information.

You needn't be a Data Scientist to perform Social Listening. A background in marketing and consumer research could be enough to do the job.

However, the disadvantage of only using these social monitoring tools to analyse consumer voice is that the output information is highly generalized hence, many functionalities like sentiment and emotional analysis, influencer detection, spam removal and others are mostly inaccurate and uninterpretable. You are inundated with massive amounts of information that you always need to summarize rather quickly. Quite a challenging task. 

This is where proper Data Science can help in making the task of Social Listening faster and accurate. However, this requires a good amount of algorithmics and machine learning -

Using sophisticated clustering algorithms we can categorize millions of conversations automatically. A contextual sentiment analysis based on classification algorithms give you a far more accurate understanding of the consumer tonality and sentiment. We could enrich the social media data with additional sources like weather data, open data, surveys etc and do correlational analysis. We can not only crunch the output of social listening tools but also crawl additional sources for information on the web through spiders. Off course, all this depends on the your client's need, scope and time.

There are multiple Data Science techniques that can be used to crunch those consumer conversations or footprints and extract invaluable insights around the brand for it's marketing and advertising strategy.

Hence, if you are a marketing or advertising expert and you are looking to analyze what your consumer says on the web and social networks, ask yourself if you would like to do Social Listening or Data Science. 

Lastly, Data Science due to it high complexity needn't be always more expensive but it's certainly more efficient, accurate and richer. Turn-key solutions are not always in your best interest.

I would be more than happy to hear your opinion on it.

Thank You.