Big Social Data For Your Organisation

Sep 19, 2016 Chloe Powell

At the moment there are more than 1 billion Facebook users, of which around 850 million are active on a monthly basis. There are more than 100 billion connections and each day gives us 2.7 billion new likes. Twitter has 500 million users and 180 million tweets are sent every day. Approximately 100 million Twitter users are active on a monthly basis. Linkedin has 200 million users and almost 175.000 user sign-up every day. Pinterest has over 10 million users, of which 97% are women. Instagram has over 5 millions images uploaded every day and Google+ receives 5 billion +1’s every day. In other words, social media is huge and drives massive amounts of data on a daily basis.

This is also called Big Social Data and more and more companies see that it contains valuable information for them to better understand their customers. Using sentiment analysis companies can understand what their customers think of their service / product offering / latest commercial etc. Even more, all the available social data can be used to perform predictive analysis about what customers want. Based on the feedback they post on the social networks, companies can obtain insights that would normally require expensive traditional research.

Hyper-Targeting

Using the available data in social media networks, it also become possible to start hyper-targeting customers. Instead of just targeting (potential) customers within a certain age, location or gender, companies can target based on actual or latent needs of (potential) customers. All based on what they mention on the social network, what the ‘like’ or ‘retweet’ and in which context.

Things went that way for a long time, but anyway the new C++11 standard was released in 2011 instead of 2009 as it had been planned. However, it was finally released and brought many additions, extensions, and syntactic sugar. It made the language much easier to code in ; the standard library was expanded with many things to make the programming process easier. A lot of new features were introduced: official support of parallel programming, initialization lists to make the code clearer, lambda functions, and much more. But there is no need to discuss all of that here because the article is not about that.

As stated by the 2nd VINT research report on Big Data, a good example of big social hyper-targeting is the company MyBuys. They offer cross-channel personalization for consumer brands and e-commerce shops. They aim to drive conversion, engagement and increase revenue by analysing individual behaviour of about 200 million customer profiles and 100 terabyte of data. All data is provided real-time, so decision-makers know what to do when.

Another good example is the Big Data startup Bluefin Labs, who have developed a social TV analytics platform to inform TV Networks and Operators about the viewers opinions in real-time. Most of the data that is collected from social media platforms is unstructured data. Traditional business intelligence cannot analyse and interpret this kind of information.

Of course, it will be a long time before programmers fully understand all the benefits of the new standards and learn to use the innovations properly. Yet these innovations are sure to have a positive effect on the quality of the code and software.

Big Social Data Pitfalls

However, you should be aware that Big Social Data is the holy grail in analytics as there are several pitfalls that could ruin the party if not addressed well:

• Not all Facebook and Twitter accounts are real. Facebook has around 83 million fake accounts and also Twitter hosts many fake accounts. Basing your business decisions on fake accounts can drive you in the wrong direction.
• Unstructured message on social media most of the time lack the context in which they should be placed. If you do not take this context into account, the data can be wrongfully interpreted. However, technology is advancing rapidly and Bluefin Labs can already place Tweets in their context based on interactions, timing and location, likes and +1’s, followers and friends.
• What is being said on social media is not always what is meant. Someone can really like a commercial because of the humour in it and Tweet about it, while the persons dislikes the product completely.
• Narcissism flourishes on social networks and especially Facebook. Everyone wants to look at his or her best and bragging happens more often than not. How seriously can an organisation take such information when deciding which target to focus on? More is being not said than is being said; so do not focus solely on what is being said.

Important Aspects

So, although there are some pitfalls with social data, it surely can help your organisation in better understanding your customer. If you start using Big Social Data, there are a few items, as discussed by the VINT report, that are important to watch:

• Is the data you use real-time; the lifespan of a tweet is approximately 1 hour and then it is gone. Social data appears fast and is gone in an instant. Real-time is necessary to have valuable input;
• Metadata helps to interpret the data, especially from blogs, faster. Ensure that metadata is included in your analysis;
• Ensure that the data is linked and integrated with other sources. The better integrated, the more relevant the data will be;

Social media gave companies an infinite source of data about their customers and buying trends on an aggregate level. All this data will help organisation in better addressing the needs and wants of their customers. Although there are still some pitfalls to watch, the available tools and algorithms are becoming better and better. So, there is a bright future ahead of us regarding Big Social Data.