Contact center (customer care) is an important aspect of any large product/service organization. It helps enterprises stay relevant in the market and enhance their products and services based on customer feedback.
With reference to contact centers, major transformations revolve around quality of delivery and the cost incurred. Some of the major challenges faced by MNCs maintaining a contact center are –
Agent Recruitment and Retention: Contact center agents are generally handpicked very carefully as one wrongly recruited agent might reflect badly on the quality of customer service. More than recruitment, retaining a good agent takes tremendous effort as call centers suffer from a high attrition rate. Companies spend more and more to come up with innovative ideas to retain their trained workforce but still are suffering majorly due to this aspect.
Increasing customer expectations: In recent times there has been a remarkable increase in customer friendly digital channels and enablers resulting in increase in competition. Growing customer expectations focused on immediate results have made it harder to retain agents. Hence, there is a pressing need to leverage technology to match evolving expectations of customers.
Workforce optimization: Contact centers are cost centers to the companies. Contact centers use innovative methods to handle the seasonal high volumes of calls during events and festivals. One of the biggest business drivers is cost reduction by work force optimization.
The objective of this article is to discuss the various levers of automation in a contact center that can be pulled to deliver a seamless customer experience.
Automation of contact centers can be done largely in two areas:
- Automation of core functional area: Core functionality of a contact center consists of receiving and/or making calls. The calls generally have an intent, like post-sale customer service, vendor support, employee assistance etc. Every intent has a pre-defined structure of conversation that the agent follows to ensure quality. The agent in a contact center is expected to capture crucial information from the conversation. This is achieved by making notes/updating a database during the conversation. The captured data is then sent to a different team for resolution based on the intent.
- Automation of quality assurance: All the calls (outbound/inbound) usually get recorded for quality assurance purposes. Quality Assurance (QA) is performed by listening to the recorded calls and rating them on quality, based on pre-defined metrics on agent performance. The most feasible approach for QA is sampling considering the efforts and cost that comes along with the volume.
Large businesses, especially in the B2C segment, are realizing that quality assurance of contact center functions is vital. It reinforces the product roadmap, offering good visibility of the product performance in the market.
The business drivers for QA of a contact center are largely influenced by cost and efficiency. Every QA agent adds to the cost. To complement the cost reduction, the quality assurance function works on samples. 1% of the total calls are sampled for manual intervention assuming it represents the larger chunk.
Automation not only replaces agents in the long run resulting in cost reduction but also enables a comprehensive check of 100% of the calls apart from its ability to handle and analyze unlimited number of calls which, ultimately, improves efficiencies.
Some features/capabilities of an automated Quality Assurance Engine for a contact center are:
- Voice-to-text – The first step is to transcribe the recorded conversation into a textual format. Analyzing textual data is relatively easier for a machine than decoding the intricacies of voice inputs. There are numerous cloud-based services like AWS, Google Cloud platform or even MS Azure services offering a transcription as a service for most of the widely spoken languages. There are also startups like Gnyani who provide speech-to-text on premise.
- Data extraction: Using Named-entity Recognition (NER), relevant data like caller name, event description, event date and time, products involved, complaints etc. can be cognitively identified and key data be extracted for further analysis. Extraction of such attributes is used to capture the summary/intent of the call. This helps get an overview on the type of calls received by the agents. These numbers complement agent training in the long run.
- Sentiment analysis – Deriving insights into the tone of the conversation is what sentiment analysis enables. Association of the tone of the callers to human sentiments like happiness, sadness, anger, fear etc. is one of the metrics to evaluate the efficiency of the conversation.
- Metrics to score an agent – An agent performance metrics is built based on an ideal scenario. Every agent call is compared against the built metrics to narrow down to a comprehensive score, for example, keyword search to identify usage of preambles, titles etc., usage of domain specific words or phrases for certain types of calls etc. Advanced NLP is used to score the grammatical correctness of the conversation.
- NLP – The Natural Language Processing (NLP) engine is used to understand the context of free flowing textual conversation. Enabling a contact center QA engine with NLP is vital as it facilitates basic understanding of the context of the conversation.
- Continuous learning: A solution equipped with AI/ML techniques, which can not only display cognitive capabilities like flagging an agent whose score falls below a threshold but also increase its accuracy over time and usage. This is achieved by a combination of supervised and unsupervised learning. A sample is first evaluated manually and fed as input to the ML model post which the solution is exposed to unevaluated calls followed by a manual verification of the generated results. The manual reviews are given as feedbacks for the machine to learn and improve its efficiency over time.
- An automated Quality Assurance engine adds value through multiple factors driving a lower cost and a higher efficiency. Some of the potential benefits by such transformations are: 100% Coverage: Quality assurance was earlier limited to a sampling approach. Insights derived out of a small sample of voice calls was extrapolated to the whole volume. Though not very efficient, this option was the best amongst the available ones. Leveraging AI/ML techniques to automate quality assurance of a contact center enables 100% validation of calls giving it a better coverage.
- Cost effective: The price of automation in the long run easily outruns the spend towards manual QA process. Moving into the era of automation, the cost of technology is going down with time giving it a stronger hold for acceptance.
- Standard and predictable: Usage of machines standardizes the process throughout making it easier to derive insights.
Some of the key outcomes that organizations expect through any transformation is cost optimization at scale with higher values delivered. With such motives we are certainly advancing towards a 100% automation-based approach with respect to Contact Center evaluation.
Conclusion
Amid the current pace of innovation, technology transformation should ideally be a part of continuous business improvement and not a one-time modification to match current market trends. We are advancing towards a fully digital contact center where human interventions will be limited to handling exceptions.