As markets become increasingly globalised, saturated and competitive in the digital age, great focus is being placed on the customer experience (CX).
And, in periods of subdued or stagnant economic growth, such as South Africa is experiencing, people slow down on their spending and as a result, it becomes harder to attract new customers. Therefore, significant emphasis is focused on reselling, upgrading, etc. existing customers.
However, there is only so much about a customer that can be discerned from their monthly account balance and credit bureau data as examples, where in many cases, whatever insights were available in this data has already been obtained. Businesses, therefore, need to look beyond the two-dimensional data available and seek other sources of information or insights that will help the business to better understand customers.
Mining the data for sentiment and meaning
Social media data that is publicly available, for instance, can be used to create a map of what is going on in the customer’s life – and provide businesses with a view of how this intersects with, and impacts on, the brand and its products and/or services. In fact, by understanding what is happening in the customer’s life today, businesses will be able to predict what the customer’s wants and needs will be tomorrow.
Mining social media data to make the most of the insights available requires a mix of technology and human touch. The reality is that sentiment analysis (SA) is a discipline that is still in its infancy. There is certainly a lot of ongoing research in this space, however, there is also significant room for evolution and improvement. And, currently achieving high levels of accuracy within SA is still quite challenging. As a result, the human touch is still imperative.
As with many analytics scenarios, the all-important work of data preparation cannot be ignored and will assist in obtaining a meaningful result. Some of the key criteria that will need to be defined for this data preparation will include: feature selection or extraction, removing noise of irrelevant data that can affect results, clear classifications of sentiment, name entity recognition (expanding on classifications), and stop words.
All the above activities are crucial to SA, and this is where the human touch comes in. There isn’t (least not yet), a one-size-fits-all model available. Many different businesses (or industries), will have different requirements here. Getting this right will involve key in-house business and process knowledge. And, leaving this up to machines, at this point, will not give businesses a SA solution that speaks to their individual needs.
Understanding the “why” of each customer
The pervasive availability of social media data has led to Business Intelligence (BI) teams being exposed to more information, which enables these teams to dig deeper, into the psychology of the customer and what is going on in their lives. BI teams can now connect the ‘what’, ‘when’, ‘where’ and ‘how’… to the all-important ‘why’ – for every transaction and engagement with the customer, as well as their ‘feelings’ about their experience.
This level of insight can give a huge increase in the relevance of the information that BI provides. That being said, there are some challenges that still need to be addressed, including:
- Being able to gain access to sufficient social data – freely available data versus private or protected data and complying with the Protection of Personal Information Bill (PoPI)
- Making sense of the data
- The cost of access to data – such as paid for media sources, for example
- The human challenge - having people in an organisation (in both business and IT) who can see the opportunities and raise the necessary funds for implementation can be the biggest obstacle.
In South Africa, we are regrettably still a bit behind in truly understanding the value of extracting insights from social media data as part of the business customer experience management and sales strategies. Social media mining is certainly a complex discipline, and such an advanced analytics strategy would be challenging to implement. However, the potential return-on-investment gains are immeasurable.
To put this into context; each day 35 million people update their statuses on Facebook
, and Facebook users (alone) generate 4 million likes every minute. But, are you tracking what your customers have to say about your business?