Call Centre Case Solution
M-Tel Call Centre data: What does it tell us?
In May 2009, Esther Ching is three months into her role as manager of the complaints section of a call centre, located in India. The call centre is owned and run by M-Tel, a Malaysian telco company. Her section handles all billing complaints and service difficulties. The main products are landlines, mobile telephones and internet access. She is required to present a report about the performance of her section, as part of the regular performance review. In the original job description her role was to “control the total costs of the complaints section while maintaining an excellent level of service to our customers likely to ensure long term return business.” The main cost is the hourly labour costs of the over 300 consultants employed in her sections. The consultants range in experience from a few months to three years. To a very large extent, consultants can be rostered on and off to match the busiest periods of incoming calls, in three hour blocks. There are roughly equally numbers of males and females, though nearly all consultants are between 20 and 40 years of age. Esther wants to obtain a snapshot of how customers react to their experience with the complaints centre as well as the efficiency of the consultants in dealing with complaints. Problem Background The complaints section of the call centre has been operation since late 2006. Prior to that date, this function was outsourced and the call centre only handled sales. Customers are often confused about the details of their M-Tel bills, especially during the first few months of their contracts. The level of confusion is not helped by the high turnover of sales staff who sometimes fail to properly explain the terms of a new contract and can send customers to inappropriate contracts for their specific needs. Customers reach the complaints section of the call centre of M-Tel through an electronic management system (EMS) that sorts the customer’s problem into several categories. The EMS is also able to have the consultant record some basic data about each customer call as well as detect data input by the customer. Three time stamps are automatically recorded by the EMS. First, the time that the customer joins the complaint centre queue. Second, it records the time that the call is answered by a human operator. Third, it records the time the call is completed. All times are recorded to the nearest second. So it is possible to calculate how long each customer waited before being answered and how long the centre spent dealing with the customer complaint, all to the nearest second.
Some customers hang up before reaching a consultant. Every thirty seconds, they are advised of the likely remaining wit time so many will hang up if the quoted time is too long. Others will initially hold and then hand up if the wait is not reducing fast enough of if it has been too long. Two important additional pieces of information are also obtained. First, the gender of the calling customer is also recorded by the consultant. The customer identity is known and on the database but the consultant records the gender by taking note of the name and voice. Secondly, and perhaps most importantly, at the completion of the call each customer is asked to rate their level of satisfaction with the company’s complaint handling procedures. They hear the following recorded message at the end of the call. You have taken up your valuable time to contact us about a problem with MTel’s billing service. We hope that the problem has now been resolved. We would appreciate your rating of you service experience with our complaints service. The consultant you just spoke with will not be aware of your rating. Please give your rating by entering one of the numbers 1 to 5 on the key pad where 1 mean very dis-satisfied, 2 means dis-satisfied, 3 mean neutral, 4 means satisfied and 5 means very satisfied.
The data. (HERE) The worksheet EMS data contains data on 1120 calls automatically stored by the EMS from the Malay state of Johor only and over the month of May. This data base is only for customers with billing complaints and excludes those customers who hung up before reaching a call centre consultant. The hold times of those 59 customers who hung up is given in a separate worksheet. It is not clear how useful these numbers are, since we do not have customer ID, service times or satisfactions ratings for these calls. The lost calls comprise only about 5% of connections in the long run and it is believed that these customers will call back at a less busy time. The data also excludes those customers who declined to give a rating. These data are available if required and comprise around 10% of calls. One could think of this as a 6th possible rating response of the customer in addition to the more informative response of 1-5. The first column headed hold records the number of seconds the customer was kept on hold before reaching the operator. The column headed service records the duration of the call in seconds. The column headed rating gives a satisfaction rating from verydissatisfied to very satisfied. This has been recoded as 1 for the worst rating and 5 for the best rating. Finally, the gender of the customer is recorded in the 5th column. Some information about the consultant is also listed, obtained by matching the consultant ID which is recorded by the EMS with their HR records.
Finally, she has some data on the number of calls handled by all complaints sections (not just Johor). The data is weekly and is measured in units of calls per minute. The centre is only open during standard business hours. The problem. Esther has the spreadsheet in front of her – a bewildering array of numbers. She somehow needs to convert these numbers into information, ideally information that leads to a better understanding of the management issues options. She first needs to organise her thoughts and writes down the following checklist.
She needs a summary of customers, the consultants and the performance. What proportion of complaining customers are males? How about consultants and their experience? How are customers rating us over all? What is the typical range of holding and service times? Which measurements do I care about here? Are any of them performance measures? Ideally would she want service times to be long or short? How about experience? Is that really her problem? Which factors affect which performance measures and how? For instance, hold time might affect satisfaction. She might also expect the best consultants to get the best results but whoa re the best consultants? Looking at the weekly data for the past 48 weeks, what is the typical call load and typical range? How many staff should this imply? Are there any patterns that require explanation?
She draws a flow diagram anticipating how each measurement or variable is likely to affect each other variable and how. The next step is to see whether there is any evidence for these anticipated patterns in the data, Esther is interested in contrasting the behaviour of males and females since this is the only customer segment information available in this spreadsheet. Consider the general proposition that male customers are less easy to please and less patient than female customers. What kinds of tables and charts might shed light on this question? Does the data suggest a clear difference or not? What other easily accessible information do you think would be helpful in analysing the complaint centre operations?