“Consumers don’t think how they feel. They don’t say what they think and they don’t do what they say”, the legendary David Ogilvy once said.

So, when surrounded by an ever growing amount of customer data, how can you really be sure you are acting on the right feedback at the right time for the right people? Here are 3 ways in which the latest innovations in predictive analytics can help you do just that.

1. Using All of Your VoC Data, Not Just Some of It

The analytics company SAS estimates that only 10% of voice of customer data are analysed (indeed our own internal research indicates it is actually far less than that, and may be a lot less than 1% if telephone calls and all social media interactions are considered). Why is this? Surely it’s critical to be able to pick out the reasons and topics which cause customers to give low and high scores or feedback, and spot hot spots of issues, unknown unknowns if you will?

The challenges are technical and economic: Text analytics can help, but it requires a great deal of skilled manpower and it requires constant refreshing, revalidation and curation otherwise it quickly becomes out of date. There are lighter-touch technologies with some domain dictionaries which can provide some directional capability and generate clusters from keywords, however this cannot be used to reliably quantify and predict issues from the rich, voluminous, textual VoC data.

Imagine if a system could analyse every single piece of verbatim from surveys, reviews and complaints with high degree of accuracy and automatically transform it, contextualise and classify it accurately?

Even better, what if when the system didn’t ‘understand’ something, it would ask the minimum input from a user to clarify? Sounds like science fiction. The good news is that it’s science fact as there are new innovations now able to do this (see Optimized Learning below).

2. Getting the True Picture from Your Surveys

When it comes to surveys, getting “yes or no” type answers, or quantitative responses such as Net Promoter Score or Customer Effort Score is relatively straightforward to obtain the ‘temperature’ of your customer satisfaction. But why did the respondents provide the scores they did? The unguided text comments, such as “tell me why you gave that score” have the richest insights and they rarely get analysed in a systematic fashion.

By using the latest in predictive analytics you can analyse all the verbatim from a series of surveys, automatically classifying keywords and patterns to transform the data into actionable insight.

For example, one of the world’s largest airlines sends out tens of thousands of surveys every month to customers of all of their services. They analysed the structured data such as CSat scores per touchpoint on the customer journey (cabin cleanliness, check in, meals etc.). However, despite having the state of the art text mining technology they struggled to analyse the free text in a systematic way and really only used verbatims to follow up the structured findings in discrete projects which involved highly-skilled analysts spending a lot of time.

The airline trialled a new predictive analytics package called PrediCX to automatically analyse and categorise the concepts and topics (i.e. not just keywords) that customers were talking about, including the sentiment, to generate an early warning system of verbatim data. The Optimized Learning technology flags any ambiguities to the users to help with manual input in order to speed up the automation of categorization.

By generating automated predictive insight from survey data, the airline was able to not only pick up the structured data such as multiple choice answers and CSat scores, but also automatically generate the reasons why their customers were happy or unhappy and get some indication on what the predictive factors were such as routes, demographics or customer segments. All in near real time.

3. Turning Complaints into Opportunities

Complaints are mostly treated as problems to handle, instead of being seen as opportunities for learning- let alone being great at converting unhappiness into zealous advocacy. Yet the format of complaints is that they are by their nature unsolicited, and can be long and contain many topics and themes.

The key aspects to improve customer satisfaction with complaint handling are: speed of problem resolution, taking a proactive approach, and the communication of next steps in the process.

When it comes to speed of response, dealing with specific complaints assigned with certain criteria can improve response rates dramatically. Whether it is based on a certain word or phrase in the subject or body of the e-mail, a particular category, organisations can apply an automatic output such as where it goes to in your team, to certain customers and more.

However, doing this manually simply involves using more and more case handlers creating additional costs for an organisation. Routing complaints automatically and prioritising by issue and category is also difficult due to the nature of complaints i.e. unsolicited, long and sometimes multi-topical. As a result, manual classification is often impossible within an acceptable time frame for the unhappy customer.

By using the latest in predictive analytics however, it is now possible to automatically classify unstructured data such as text and provides an early warning for issues that need resolving fastest. What more, this all happens in near real time.

Conclusion

So, could predictive analytics be the answer to getting a true 360 view of an organisations’ VoC data? Well, if you want it in near real time, without losing valuable data along the way and without requiring 80% of your data scientists time spent transforming text data, it’s definitely worth checking out the latest software releases out there.

Interesting Links:

Post Views: 1335