The Enterprise Customer Care Challenge

A customer care division is the heart of the Enterprise, and especially in service organizations. A typical customer care division is based on its Call Center, in which the service reps are playing an important role when answering to customers’ questions and complaints. 50% of the Enterprise overall costs are related to customer care, of which the manpower cost takes the majority part.

Enterprises are trying to reduce the enormous operational costs by assigning the calls to IVR, and by encouraging customers to consume the service from their official websites (Web Self Service, FAQ). However the analysis shows that the savings in the Call Center is not sufficient.

The basic analytical structure of the telecommunication call center is listed bellow:

From this analysis it is clear that the enterprise wishes to transfer customers to the IVR as its call’s cost is 85% less than a rep call’s cost. The customer care goal is that only minimal enquiries will arrive to the rep desk.

Another major problem is the level of service which is viewed from the following aspects: customer waiting time, lack of service rep’s experience (changing positions every 12 months), and the fact that full service is provided only at the normal business hours. At night and in weekends only a partial service is provided.

Semantica is offering a solution for improving both call center issues, the operating costs and the level of service.

The problem of Self Service websites (the alternative way for getting support)

Customers are expecting fast access to self services websites. They would like to access a site, type their enquiry and get an accurate reply right away. The statistics shows that more and more customers prefer to get online support rather than calling the call center and waiting a few minutes for help.

The advanced Self-Service websites offers a lot of information. In many of them the customer is also offered to type his request in free text. However the implemented technologies which are based on keywords search are not accurate. In most cases the customer is offered to read dozens of alternative questions, for finding the predefined question that fits his request. In many cases after going over all the questions there is no question that fits the customers request, and the customer is left without a reply.

A search procedure that is based on a combination of keywords usually shows poor result (less than 35% accuracy).

E-Rep Solution

E-Rep receives a free text query from customers. It analyses the text using Semantica’s innovative Natural Language Processing (NLP) technology and provides a highly accurate results.

E-Rep understands the full meaning of a complicated sentence by identifying the sentence semantic structure like subject, verb, arguments etc. Examples of typical enquiries:

- “How can I operate the alarm feature on my Nokia N800“ – a question with 1 arguments.

- “What is the price for calling from France to the UK at the coming weekend“ – a question with 3 arguments”.

E-Rep is based on the most advanced Natural Language Processing techniques called Semantic Role Labeling (”SRL”). SRL performs efficiently a profound analysis of complicated sentences in written or spoken free language. E-Rep achieves a recognition rate of more than 85% and in cases when the system “suspects” that it didn’t understand the free text, it asks the customer a directive question to provide an accurate result.

Currently we support analyzing text from textual media (email, Self Service sites, SMS) in the English and Hebrew language.