Call Centre Analytics
Call centre analytics is a solution for text analyses of calls (Voice to Text), answer rate prediction, predictive models for targeting or a capacity model to improve the success rate of call centre calls.
Deloitte has acquired the knowledge of all critical factors affecting the success and efficiency and effectiveness of calls and other customer call centre activities, integrating this knowledge into our unique call centre analytics solution. Our solutions are customised and always adapted to the client’s specific needs.
The solution makes it possible:
- For call centre directors to sleep soundly at night;
- To increase the availability of call centre employees;
- To increase the success rate of calls;
- To speed up the solving of incoming calls;
- To decrease call centre costs;
- To increase the profitability of call centres;
- To optimise call time, call duration and the quality of communication; and
- To optimise the capacity of operators and the transfer of knowledge.
We have assisted the internal call centre of a major Czech bank in doubling the sales success rate of its incoming calls.
Call centre analytics may comprise one of the components below or a multiple of them:
Call centre audit
If we are to truly help a call centre, we first need to identify in which areas this is actually necessary and in which it is not. If the client is unsure, we recommend starting off with a gap analysis, which will assess the call centre’s level in the following areas:
- Motivational chart;
- Reports; and
We will provide a comparison with the best practice in these areas and identify minor opportunities yielding benefits fast without the need to interfere in the IT infrastructure, and we will describe and design the implementation of major pilot events.
Would you like to acquire a general grasp of all the things that are said during the day at the call centre, how customers respond and what it means for you KPIs? We can do this! Thanks to the modern Voice-to-Text technologies, we will convert speech into the textual form, measure the call time and duration, and mark when the operator was speaking and when the customer was, including other call features (silence, speech, answers, hesitation, interruption).
We will subject the textual transcripts of calls to textual analysis at the level of words (basic level: morphological analysis, lemmatisation, sentiment analysis) or at the level of sentences and sentence units (advanced level: sentence analysis and the analysis of natural language). We will put the information in context with the target KPIs (eg the service to sales conversion) and using machine learning methods and big data technologies, we will identify hidden interdependencies and the factors of success or failure. The next stage will involve scouring the swathes of data for the essential and important information to be reported, and creating a well-arranged interactive dashboard that will make it possible to put the new insights into practice. The last stage involves implementation, whereby the process of the daily calculation of underlying data is created – one that must be fast and reliable. This will again entail making use of big data technologies, without which the objective could hardly be achieved. Insights acquired during the actual project include the following:
- Conversion is the highest in calls where the customer was silent for a long time.
- Conversion is higher if the silence before the operator speaks is only short.
- Conversion is higher if the customer’s shortest speech is very short (eg yes, no).
Right Party Contact
One of the key parameters of outgoing call efficiency and effectiveness is Right Party Contact (RPC). RPC has a typical curve during the working day and week, yet the peak RPC times coincide with the busy hours of operators. As such, the classic trade-off between efficiency and effectiveness and capacity planning is addressed.
Yet, from the perspective of the customer, right party contact is a highly individual matter. Customers may be divided into early birds, those who deal with calls on their way to or from work, and those who are active mostly in the evening and at night. To pick the right time for making a call when the selection is random is almost impossible. How much would right party contact increase if you contacted each customer at the right time?
The RPC model predicts right party contact and call duration at the level of individual customers depending on time. Using customer data, it is possible to create relatively sound and accurate models these days. Looking at the RPC curve, you can easily determine its peak, plan a call and determine the likelihood of having the call answered. But this is just the beginning of the story. The task is to take all customers and their curves, take all the operators who will be available and plan tomorrow’s outbound calls so as to maximise the total RPC in utilising the capacities of all available operators to the optimal degree. This is a mathematical task from among operating research that we are able to solve. The result is simple: a detailed plan of outbound calls quantifying the anticipated right party contact, the number of customers reached, the utilisation of operators and other operating metrics, if any, taking the form of a well-arranged dashboard.
Predictive models for customer targeting and retention
The success of the call centre’s outbound calls (cross sell, upsell, activation, retention) depends not only on the quality of the call centre as such, but, to a substantial degree, on who you are calling to, when you are calling and for what reason you are calling. We offer a series of analytical solutions to this broad range of issues, such as Propensity to Buy (PtB) models (refer to Customer Targeting), Propensity to Churn (PtC) and Propensity to Save (PtS) models for retention (refer to Customer Retention), or the Customer Lifetime Value (CLV) model (refer to CLV). If you are an enlightened call centre director and your colleagues in marketing have yet to wake up, do not hesitate to ask for these tools. They are currently at the level of best practice, yet they are slowly becoming the industry standard.
How many people are we going to need for inbound? What is the likelihood of violating the SLA if there are only four of them? Shall I call in reinforcements for outbound when the best retention operator is on holiday? For call centre managers and team leaders, these and many other questions are the order of the day in planning capacities. These kinds of issues may be addressed one by one based on experience and an expert estimate, but also analytically all at once. At Deloitte, we believe that analytics-based capacity planning yields better results.
Our call centre capacity model searches through all possible capacity plans to find the one that best meets the required characteristics, such as:
- Minimal dropped-call rate (inbound RPC);
- Maximal right party contact (outbound RPC);
- Meeting of requirements for outbound;
- Maximal operator capacity utilisation; and
- Appropriate allocation of skills to individual calls.
The resulting optimal plan needs to be effectively delivered to all the call centre’s employees. Based on our experience, we recommend integrating and extending the existing system so as to preserve the compactness of the user interface for operators and not to implement a new application.
If set correctly, analytical capacity planning will take the burden of the everyday routine of complex decision-making away from managers and team leaders, enabling them to concentrate on coaching and quality improvement, or they will simply be able to get a good night’s sleep.