Contact centre Performance and Key Performance Indicators A white paper by Independent Perspective Contact Centre Performance Data design needs to be revisited to ensure that it supports the need to manage and improve in today s environment. Customers now expect to be able to call or email and experience instant gratification of their needs through an answer or supply of service. This paper is the third of a series prepared by or for Independent Perspective Australia to raise and examine issues of relevance to our friends and customers. Each of the papers is intended to put a view on a particular topic which hopefully will create discussion and awareness. Contact centres today have become a key part of the distribution and servicing of products and services for most companies of scale. Customers now expect to be able to call or email and experience instant gratification of their needs through an answer or supply of service. Of course, providing this level of service is expensive and requires a highly sophisticated approach to resource management and allocation. It is critical under these conditions to get the highest levels of resource performance possible. In this pursuit, the typical contact centre manager has to deal with a vast array of performance data. Our research shows that Contact Centre managers feel that they have too much data already. Unfortunately, when asked, they are often unable to identify which of the current data set it is that improves performance. In today s contact centres we at IPA have found that most of the performance data that is available is sourced from the telephony platform in use and that, although there is a rich stream of data available, the most frequently used is that which directly relates to accountability upwards in the organisation. In other words, contact centre management does not analyse and evaluate information based on what would make them a better function but rather what exists. In this paper we will raise a few issues with popular measurements and hope to start management thinking about using performance data to create a better function and constructing performance measurements that sustainably support that goal. In other words, contact centre management does not analyse and evaluate information based on what would make them a better function but rather what exists. About Contact Centre Measurements Contact centres are by definition aggregations. They may have 2 staff, they may have thousands. An effective measurement methodology for them needs to be able to adapt to the appropriate number of agents, their physical location and the variety and complexity of contacts either taken or made. Add to this mix the need to link not only with core applications but also back office, regulators and other stakeholders and it is easy to see the difficulty in making metrics accurate and relevant to the purpose. In order to understand where contact centres are today, we should endeavour to get an historical perspective. Consider a contact centre for a major organisation in the early 1990 s. It would have been charged with a simple customer service role. Staff would have been using legacy software applications and mostly looking at green screens.
Their authority to make real changes for the customer would have been limited by the need for paper authority. Wait time targets would largely be a thing of the future and most of the staff would have seen the job as an entry level into a real job within the larger organisation. In today s contact centres managers face challenges that derive from the new and critical role that they play in the success of their organisation. Let s look at some of these challenges: Mission Criticality Today, contact centres are a key part of the distribution and service strategies of most large organisations. Contact centres must sell the product, give up to date advice of a transaction status and produce data for a variety of stakeholders. Indeed, customers now see contact centres as the primary connection to the company as physical branches and mail are increasingly a thing of the past. Metrics to ensure that these various performance challenges are met must now reflect the diversity of owners and needs. De-Aggregation The contact centre manager of today is often faced with a physically de-aggregated group. This may be two groups in the same city or over several cities or even countries. This presents special challenges in the delivery of service to customers and stakeholders. The fact is that even though the products and services are the same, the demographics and culture of staff and the industrial relations landscape may be different leading to differences in costs per transaction. Key Service Level Performance Indicators As a channel that is core to the delivery of the organisation s products, contact centres now need to deliver to specified quality, financial and time related performance indicators. This is especially relevant where there are additional statutory controls such as financial advice. Customers now expect long hours of availability, almost instant connection and an ability of the contact centre staff to resolve issues during the call. This means that new performance indicators are required measuring fulfillment of these requirements. Cost as a competitive advantage As the role and consequent size of contact centres has increased within organisations controlling and seeking a cost advantage has become critical. So a contact centre must deliver more for less and measurement must deliver insight on costs and how to continually reduce them. In order to meet these challenges new management tools and methods are required. Management must be far more tactical and data based. As we will demonstrate in this white paper, that data must be sufficiently granular to tell the entire story. Delivery of performance data has largely been the responsibility of telephony and technology companies. These companies delivered the PABX, then the ACD and CMS applications. Unfortunately these vendors have demonstrated limited capacity to deliver measurements that support the goals of specific users and in a manner that leads to action. Most Common Key Performance Indicators It is the contention of this paper that most performance measurements currently used by contact centres have been created and driven by two factors. The first is the availability of measures from the telephony platform vendor and the second is the constraint imposed by availability of data from existing sources. It is certainly a conundrum that a contact centre manager has a massive amount of performance data available but the actual data required is often unavailable due to the cost of collection. We will consider the most common data usually available and currently in use by most contact centres and see how it holds up in today s environment and then go on to suggest a better way to build performance measurement. It is certainly a conundrum that a contact centre manager has a massive amount of performance data available but the actual data required is often unavailable due to the cost of collection.
Traditional Measurements 1) Average Handle Time (AHT): AHT remains the gold currency of the contact centre. It was the first measure developed and reported on by telephony platform vendors and it has never lost its grip. We would contend that it is time that it is replaced for many companies with more suitable measures and we will illustrate this with a simple example. Name Gender Height (Cms) Weight (Kgs) Peter M 195 125 John M 120 65 Paul M 170 80 Barry M 175 70 Mary F 160 50 Average M 164 78 Table 1: Example of average limits This is an amusing but real example of the severe limits of averages in real life. Averages are no more or less valid than any other measure but they operate, to be effective, under some rules. All data elements need to be of the same class. For examples Males and Females are of the same class of humans but they are not divisible as class subsets. In the table above the average gender is male but this is meaningless data for many purposes. In addition data elements have to comprise a fairly narrow range. Averages also have the theoretical limitation that while they represent a data class they need not be a member of that class. Therefore in the above table there is no 164 cm male who weighs 78 kgs. In fact there is no one even close to that. Similarly Average Handle Time can encompass a range of dissimilar and incompatible elements making it a dangerous index to use in the management of performance and people. For an average to be of use in any situation as a basis for action, the data set has to meet a number of conditions. Among these are: A range of values that are narrow to the average so that the average reflects the majority of value instances closely or a set of value instances so large that variance to the average matters little to the outcome A set of values that are functionally consistent (e.g. for many purposes male and female are not divisible and therefore not members of a valid sub set i.e. there can be no average between these values for most purposes). What this means in task measurement is that if two tasks are an either/or alternative they cannot be used as a data element in an average. A set of values that make sense when grouped. For example grouping discretionary activities such as meetings with a processing task would not make a lot of sense yet often employees are judged on output over a day which includes discretionary activities such as meetings. A sales person forced to carry out administration most of the day would be entitled to feel that output per hour worked was a poor KPI. 2) Total Output It is astonishing how many otherwise sophisticated companies base their performance measurements on a variation of total output. In these days of quality and service management companies will still regard output and backlog as key measurements in isolation from other factors. Largely this is because it is usually an easy number to obtain from existing core applications and often it is because other factors are considered unable to be measured.
While output is what pays the bills as they say, it is a measurement that should be used with caution when managing staff. It can be impacted by so many factors, some positive and some negative. If we are measuring sales conversion rates for example we need to make sure that we take into account the competitive aspects of the product as they change over time. A substantial price discount or feature market lead for example will substantially impact on results. On the other hand management failures can have a negative influence leading to lower sales. Output therefore should not be forgotten but needs to be placed in context with other factors. 3) Customer Wait time This is one of a variety of service measurements. It is easy to get from a telephony platform and has therefore become ubiquitous. While it, like many other service measurements is extremely important it also must be used with extreme care. An easy way to manage customer wait time is to increase contact points (or staff) and likewise all other things being equal, a reduction in contact points will lead to an increase in wait times. Reductions in wait times can also come about sustainably through contact avoidance, a topic we raised in our last White Paper. Getting to the reason for the call and fixing it can have a positive impact of up to 40% in reduction in contacts. Even in a sales environment this can work as the goal is to convert the sale on the first contact rather than as a result of multiple contacts. Working on better Performance Measurements In the following table we consider a contact centre of 2 agents with 4 call types and a break period. Most of this information would be available from modern automated call distributors (ACD s) although most will not be able to manage standard times or make comparisons. John Barry Rostered Start time 8.30 8.30 Actual Start time 8.29 8.40 Tasks Task Std Time Number of Tasks Actual Time Taken Mins Total Std Time Achieved Mins Number of Tasks Actual Time Taken Mins Total Std Time Achieved Mins A 5 10 50 50 8 50 40 B 2 50 100 100 46 100 92 C 20 15 140 300 5 200 100 D 10 15 110 150 1 25 10 Break 20 1 20 20 1 30 20 Rostered Finish 4.30 4.30 Actual Finish 4.40 4.25 Rostered Mins 420 420 Actual Mins 431 405 Std Mins Earned 620 262 Productivity % 144% 65% Table 2: Contact centre of 2 agents with 4 call types and a break period
We can see from Table 2 that both agents were rostered to start at 8:30am. In this case John commenced at 8:29 and Barry at 8:40. This type of roster adherence is tolerated at most contact centres largely due to the fact that the management and staff do not realise the cumulative or wave (see insert below) impact of not having the rostered number of agents available for calls at shift commencement. We can also see that John is taking half of the standard time for Task C and significantly less than standard for task D. Barry on the other hand performs poorly on those tasks but almost as well on task B. Overall productivity therefore which is once again an average, just like AHT, only tells a part of the story. Analysing variance to standard by task produces an insight as we illustrate in table 3 below. Task John Barry Actual average time Variance to Std (mins) Actual average time Variance to Std (mins) A 5 0 4-1.0 B 2 0 2.1-0.1 C 9.3 +11.7 20 0 D 7.3 +3.7 25.0-15.0 Table 3: Analysing variance to standard by task Again, we stress that this is an extremely simplified example yet is also one that is representative of a pattern we observe in many contact centres and over many thousands of cases. In this example, Barry is clearly not as good as John in call handling. What stands out however is that in call type B Barry is very nearly as good as John. There may be many reasons for this but at the very least this table makes clear that if we are able to present Barry with more of the B calls and less of the others we will significantly improve his productivity. At the same time if we are able to present John with a higher proportion of D calls we will see super improvements in overall call handling. This process is known as skill routing. It is the most effective way to identify and improve efficiency in any organisation. By this process we have in this small example created the opportunity to increase efficiency by 23% overall. That is, we have John process all of the D and E calls and Barry all of the B. In theory the profit from this process is endless, as long as we have the measurements available to continue to drill down to the type and content of call and the corresponding ability to present the agent with a specified type of call. the presentment of the call types will be supported by most modern ACD/CMS systems but must also be supported by research into valid call categorization and the resulting impact on customer experience. One of our clients identified 221 significantly different call types. In this case management by simple ACD system software was a fantasy. Most cases however, fall in between example one and the most extreme but it is also true that in a modern contact centre ACD-based metrics and applications will not be up to the task of producing this type of information with the granularity required. The Wave In a high volume contact centre most managers tolerate or accept a degree of tardiness in their staff. This often manifests as the difference between arrival and commencing work. If the volume of contacts is high this creates a backlog of calls that cannot be cleared in the next period. Customers will abandon and try to reconnect later. Thus tardiness creates a Wave that can adversely impact performance for an entire shift. This type of analysis can be used for the tasks of selling and sales data. Too often an aggregate measure is used to the detriment of the organisation. In Table Three above we see that John, Barry and Susan have different selling experiences but which is the best for the organisation? John has a conversion rate of 100% compared to significantly lower performances by Barry and Susan. Susan however is clearly the best value for the organisation even though she has the lowest conversion rate. Her total sales are $2,400 or 140% higher than John due to the combined factors of making a lot more calls and a higher value per sale. Of course this example is only valid in a situation where there is a high volume of potential sales calls. In a high value, low number potential environment John s conversion rate may be the key attribute. The example is used to illustrate the need to drill down to the key determinant factors before making a decision.
John Barry Susan No Sales Calls 10 20 30 No sales 10 15 20 Conversion rate 100% 75% 67% Value per sale $100 $100 $120 Total sales $1000 $1500 $2400 Table 4: The importance of conversion rate What is the Best Key Performance Indicator to Use? The best measurement to use is the one that will drive the business in the direction that supports the organisational strategy. In most cases this strategy has multiple aspects. Cost per Call is always a driver but AHT as a measure may lead to lower customer satisfaction or lower sales. Therefore Cost Per Transaction may be a better test but even so it must be measured against service and quality levels. In an earlier White Paper we looked at contact centres and Mazlow s hierarchy of needs (please contact us if you missed out on a copy). Our conclusion was that the position of the contact centre staff on the hierarchy had a major impact on their ability to deliver certain outcomes. For this and other reasons a contact centre can have a good cost per transaction but a poor outcome. Cost per transaction can, however be validated externally and is therefore remains good overall test of the success of the contact centre not only against other contact centre operations but against other forms of service and delivery as well. Cost per transaction is not relevant to individual operator performance however. Operator performance must be measured by an indicator which has the following properties: Measurements must clearly and accurately reflect the task undertaken. Nothing will produce a worse outcome than a performance measure that does not take account of the difficulty or complexity of the task undertaken. Time based measurements will not measure empathy or quality so measures have to be broadened to include these factors. Quality is easily available from applications like Nuqleus 3D and empathy can be gauged by customer experience surveys. How Do We Develop the Right KPI s? It commences with an understanding of the customers and the processes that are important to the success of the organisation. It continues with a clear strategy of features and benefits to be delivered to the customer and ends with an application that will allow you to accurately and conveniently measure and track all of these factors. Understanding customers and processes is a reverse engineering process. You need to work out what it is you want to deliver and under what conditions and build the process back from that. In this process of specification you should avoid averages unless you are confident of their applicability and validity. If you are missing information critical to the process you should mark it for later reference rather than ignore it. You need to get the right measurement application in place and train your staff to use it well. Contact centres, as we noted at the beginning, have lots of measurement. They just don t often have the right ones. Lastly, you should continue to analyse and evaluate the impact of your measurement on the process. Remember - you get what you measure.
About our White Papers Independent Perspective has been providing consulting services for operational improvement to large companies for over 20 years. This paper is the third of a series prepared by or for Independent Perspective Australia to raise and examine issues of relevance to our friends and customers. Each of the papers is intended to put a view on a particular topic which hopefully will create discussion and awareness. We hope that you enjoy this white paper and find it professionally rewarding. We would welcome any feedback from you that would allow us to develop it even further. Please contact us if we can assist you in development of your contact centre effectiveness on 1300 90 20 90. eg solutions plc is an operations management software applications vendor. Our software provides real-time, historic and predictive Operational MI. When implemented with our training programme for managers and team leaders to use this intelligence, we guarantee improvements in operational results in short timescales.