Coincidentally, a couple of weeks ago I read on the news about a very well known low cost airline company being ranked in the worst position when it comes to customer service in a survey run by a consumer magazine. Within a few days I received an email from the same airline company about completing a customer satisfaction survey that is linked back to my personal profile. Clearly it appears that they decided to act on the findings and I was contacted. The email triggered my curiosity and I did a little bit of research around their future plans. Obviously they are planning to listen to their customers more. They are setting up a team to reply to complaint emails and they are establishing a twitter presence.
Survey data, complaint emails, twitter tweets: they are all results of different means that consumers use to share their opinion or to express their satisfaction/dissatisfaction about a product or a service. Undoubtedly they contain information that is very insightful but challenging to analyse.They are called unstructured data. Unstructured data are text-based and do not have a rigid structure or form. This lack of structure makes them difficult to organise and analyse, as they cannot be processed using traditional analysis software.
The other challenge around unstructured data is their volume. In 1998, Merrill Lynch estimated that around 80-90% of useful business information starts off from unstructured data. Since then the volume of unstructured data has been increasing at a rate of 10-50 times faster than structured data. IDC foresees that the amount of digital data will grow 40% to 50% per year. Figure 1 shows that by far the biggest proportion of digital data is unstructured information.
Figure 1: Total Enterprise Data Growth 2005-2015
How can organisations exploit these data and quickly capture the insight they contain?
Text analytics technologies enable companies to mine this vast amount of free form information. They do not only search for key words but they can also understand the feeling around what has been said or written. They can pick up sarcasm, irony, dissatisfaction, which means that businesses can better understand customer behaviour. By using text analytics technologies the unstructured information can be converted into traditional, structured data. This means that these data can then be easily combined with other sources of data, such as transactional information, web analytics. Without a question this would allow organisations to comprehend and improve their customer experience.
Most of the data mining technologies such as SAS and IBM SPSS offer a text analytics component. Figure 2 shows an example output of complaints data which has been analysed and converted into structured information.
Figure 2: Output example of text analytics
We will stay tuned and see if the low cost airline company will improve their customer service by analysing the data that they will obtain through their twitter account, surveys and complaint emails. And who knows, maybe one day we will start receiving personalised emails from the airline companies based on what we tweet and where we usually travel rather than generic bulk emails that remain unread in our inboxes.