Close this search box.

GUEST COMMENT Unstructured social data – the building block of big data success

This is an archived article - we have removed images and other assets but have left the text unchanged for your reference

by Tom Smith

The legacy of the past year’s Big Data buzz is clear to see. The majority of marketers today are fully engaged with the term, and many have even mastered the difficult balancing act between ambitious theorising and practical application, successfully incorporating Big Data usage into their business strategies. The repeated refrain of “data is nothing without insight,” has become familiar ground for retailers who are now savvy to the importance of harnessing readily available information, both internal and external to their businesses, and using it to understand their customers better.

It seems, however, that a different side of the Big Data coin remains largely unexplored. The use of unstructured data, collected from sources such as social media, internet forums and LinkedIn, is still in its infancy within the commercial world, but is rapidly evolving along with the technology created to measure and refine it. This type of data has the potential for rich and powerful insights and has several discernible advantages over the kinds of information marketers are now accustomed to dealing with.

Social media is the richest source of up-to-date customer insight in the world of unstructured data; it can act as a real-time window onto how people are feeling, behaving and communicating with one another. Most major agencies and brands have been using some form of sentiment analysis for the past several years to ‘listen’ to customers by monitoring billions of social media conversations and examining linguistics to find out how customers are feeling toward a particular product or service they provide.

However, the usefulness of measuring sentiment is itself a highly debated area; although it can provide a limited degree of insight into whether or not customers are happy, sentiment is generally a reactive approach to using social data, and as such can only tell marketers about things that have already happened. If consumers are unhappy with something, there is very little sentiment analysis can tell marketers about how to avoid the same mistake in future. The same goes for if the sentiment is very positive; for example, if Burberry produces a new range of coats and customers rave about it on social media, it is very difficult to pinpoint which aspects of the product campaign people are really responding to – and even harder to use any of this information to improve on the next stage of the marketing process.

The next step is for marketers to begin using social data to predict business outcomes rather than just ‘listen’ to what customers are saying. One possible way to do this is to use sentiment to be active (instead of reactive) during a product launch process; global software giant Microsoft and US telecommunications company Sprint are two notable examples of companies that embraced social data in this way in 2013.

Microsoft wanted to learn how best to effectively respond to a crisis to limit the damage done to the brand if such a situation should arise. Using sophisticated social data analysis technologies, it examined recent cloud outages that had been in the press, including Google, Hotmail, Skype and Amazon, and looked at how consumers dealt with the various meltdowns by turning to social media. The result was a new action plan put in place in case of a crisis, which delivered straightforward communication with regular updates across multiple channels to help customers recover quickly. When Office 365 suffered an outage three months after the data analysis, Microsoft was well-prepared with an action plan, and the impact to the consumer was minimal as a result.

Sprint was looking to add a new business-to-business mobile phone to its range, and wanted to find out more about how its brand was perceived in the B2B community. It used detailed analysis of social data to identify that customers needed to be educated and reassured more with regard to infrastructure and IT security issues, as well as the fact that people tended to buy ‘business’ phones in a very different way and with different motivations than they would buy a phone for personal use. Armed with this new information, Sprint was able to re-evaluate and improve its product launch strategy to fit better with the requirements of business customers.

In stark contrast to the sorts of customer data gathered through surveys and market research activity, social data is free and easily available to anyone who would like to use it. Of course, to make sense of the data itself can cost money, as social data needs to go through a process of filtration in order for it to be ‘clean’ and therefore useful – this can involve using specially developed technology, or a team of human analysts, or both. As a result of this necessary extra step, unstructured data is still a rather more costly, labour-intensive process than, for example, using Google Analytics. This initial premium of time and money will almost certainly become cheaper and more efficient as time passes and more is learned about the possibilities; making the results well worth a little extra effort.

Another major benefit of social over traditionally-harvested data is that it is real-time, and therefore can allow marketers to react quickly. Under the current structure of most organisations, analytics reports tend to be sent out to marketers once every month or so – but this is far too slow to keep up with customers’ fluctuating demands, as changes can only be made once the data reports are delivered, and these may well be out-of-date by the time they reach the people that matter. Social data, however, can show marketers exactly what people are saying from one day to the next, making slow reactions a problem of the past.

Perhaps the most important advantage of social media is that it is organic, and can be said to accurately reflect what people really think. A huge amount of money is poured into creating surveys and feedback forms to give the most realistic set of results. Unfortunately, no matter how much effort is put into creating an unbiased question, the answer is always potentially impacted by the fact that there is an incentive to fill out the survey itself, the customer is giving the expected answer rather than what they really think, or could be pressed for time and are completing it in a rush.

None of these problems apply to social data; with millions of tweets posted every second, this abundance of information is volunteered freely and without bias. What’s more, customers tend only to post on social media when they have strong views either for or against a brand – marketers can rely on this extremity, as there is very little grey area, and customers are likely to be either powerfully advocating a brand or openly asking for help or support with their problems.

As marketers learn more about the potential of making use of social data, the world of commerce can expect to see big changes in the way organisations interact with and learn from their customers. The relationship between brand and consumer is likely to change in a fundamental way as businesses begin to trust social data in the same way they would trust financial data; letting the customer’s voice actively influence crucial business outcomes instead of just serving as an added way of ‘understanding’ people.

As with any nascent concept, unstructured data currently demonstrates many challenges as we work towards catching up with the pace and possibility of social media – but the ultimate goal of providing customers with better content, better recommendations and better products remains the same.

Tom Smith is product marketer at SDL

Read More

Register for Newsletter

Group 4 Copy 3Created with Sketch.

Receive 3 newsletters per week

Group 3Created with Sketch.

Gain access to all Top500 research

Group 4Created with Sketch.

Personalise your experience on