For the last few years, there has been a discussion in customer analytics around the possibilities of using Social Media (and other verbatim) data. Organisations find that insights from Social Media data do not always coincide with their business information. This is exemplified by research done for a retail bank where many customers commented negatively on the counter queue times; however measuring the actual queue time (operational measurement) revealed that the particular bank had the lowest queue time in the market.
The discrepancy between social media data or general verbatim and business information is due to a number of reasons including:
1) Customers’ perceived personal gains with the comment
2) Time lag between the actual experience and the assessment
3) Over-/underestimation of emotions in the social space
1. Customers perceived personal gains
Customers’ information is likely to be biased due to their perception of personal gains. Think of the following example: a family vacationing at a tropical resort updates their social profiles with comments of a fabulous vacation to be viewed by their friends & peers. Their updates only focus on the positive elements and are often slightly exaggerated as the personal gain is to portray “the perfect family holiday” to friends and family (positively biased). However, after the holiday when the resort asks them for their customer feedback they are likely to provide a slightly more negative assessment (compared to the social media updates) as they will be prompted for specific features of their stay – maybe the lack of variety in the food, or the uncomfortable beds.
Sharing personal information is not only distorted towards personal gain but it can also restrict information. For instance, while applying for a loan or mortgage, we are willing to share all sorts of information also completely unrelated information in order to build character. However, if a research company calls to assess our satisfaction with the mortgage or loan application, it is unlikely that we will share personal information as the personal gain at this stage is far less.
2. Time lag between experience and Social Media posting.
It is a well-known fact in the analytics community that the longer the time lag between an experience and an assessment, the more inaccurate the assessment gets. Therefore Social Media seems to be an excellent way of collecting immediate customer feedback. However, unless you can pinpoint the experience (a trade promotion was launched the day a comment about it was posted) the Social Media results can be distorted based on a hidden time lag.
There are a number of ways you can counteract this bias:
- If you can pinpoint the time between an experience and the assessment, you can then down-weight responses that have a longer time lag
- Search and remove comments that have specific time references such as “last week I tried the new shampoo”
- Exclude re-tweets or comments posted as replies to someone else’s assessment as these second line comments can be old experiences that have just been prompted by the original comment.
3. Suppressing negative emotions
An interesting research by Jordan et al. (2011) found that people are less likely to discuss their negative feelings in public. The data showed that 29% of the bad experiences occurred in private compared with 15% of the good ones. Further, the research also found that people deliberately concealed negative feelings 40% of the time. This is a very important element in understanding what emotions are voiced on social media – people in general are less likely to voice negative emotions in this social space compared to a less social environment. This means that companies need to be mindful of using this information together with other sources such as customer service data, complaints or simply other survey data where responses has been provided in a non-social environment.
Is the Social Media data completely useless!
Of course not! In the example of discrepancy between actual and customer feedback on counter queue time, digging further through the comments provided an understanding of some underlying problems that the operational measurement did not take into consideration – it showed that in order to reduce counter queue time, customers were pushed to back office counters for more advanced transactions, which did not have a queue management system in place. The comments prompted a root cause analysis here.
Social media or other verbatim data is very useful as long as companies understand the assumptions on which it is provided and take corrective actions accordingly. The danger is using the data in its raw format without considering the potential bias alongside other types of business information and thereby providing conflicting messages to the organization.
All in all, Social Media data can be a valuable source of information to illustrate already identified problems, measure effects of marketing activities, used for root cause analysis etc. We just need to bear in mind that social media comments are provided in a social environment where the customer is aware that the whole world is listening and if a slight exaggeration can make it more interesting – so be it.