BENCHMARKING AND DEFINITIONS WITHIN CONTACT CENTRES
SECTION 4: AN ALTERNATIVE APPROACH TO BENCHMARKING
The use of ratios for comparing performance can be problematical. For example, to evaluate the operational efficiency of a school, an accounting ratio such as cost per unit of pupil achievement might be used. A school with a high ratio in comparison with the ratios of other schools would be considered less efficient. But the higher ratio could be due to a more complex mix of transactions. For example, a school dealing with a high level of cultural diversity may require more time to create pupil achievement than a school that is not. The problem with using simple ratios is that the mix of outputs is not considered explicitly. This same criticism can be made concerning the mix of inputs. For example, some schools may already have at their disposal extra levels of learning hardware and software that could affect the cost per unit of pupil achievement because fewer teachers may be required.
Broad-based measures are highly relevant as overall school performance measures, but they are not sufficient to evaluate the overall operating efficiency of the school. For instance, one could not conclude that a high achieving school is necessarily efficient in its use of personnel and other inputs. A higher-than-average proportion of able young people could be the explanation, rather than cost-efficient use of resources. Hence we need a more comprehensive approach to evaluating relative performances.
DATA ENVELOPMENT ANALYSIS (DEA)
Data Envelopment Analysis (DEA) is a linear programming technique developed by A. Charnes, W.W. Cooper, and E. Rhodes (1979, 1981). Its attraction for benchmarking research is its ability to convert multiple inputs and output measures into a single comprehensive measure of efficiency. As originally developed, DEA compares the input/output performance of individual productive units (companies or businesses) or Decision Making Units (DMUs). It evaluates relative efficiency between the specific DMUs being considered. DEA does this by constructing an empirically based 'frontier' from linear combinations of all the other DMUs in the sample. Hence, an efficient, or best practice organisation can be defined as one which is able to produce the same level and mix of services as other DMUs in the analysis, but which uses fewer inputs. Alternatively, the efficient DMUs can be defined as the ones that produce a higher level of output for the same levels of input.
The DEA methodology was developed originally to evaluate relative performance in non-profit and public sector organisations and is being used increasingly in private sector applications and in the service sector in particular. The literature includes numerous studies in multi-site units in banking, education, dairy farming, forestry, local government, police and health care. Its primary advantage is that it can be used to compare similar service providers, computing efficiency ratings based on the ratio of resource inputs to outputs. The method identifies efficient units, which represent 'best practice' and areas for improvement in those units identified as performing less well. By considering the relative efficiency of units with differing characteristics, we can disaggregate the impact of a range of inputs in relation to achieved outputs. Hence as an evaluation and benchmarking approach DEA can have considerable advantages over conventional ratio based methods.
4.1 COMPLIMENTARY METHODS TO MAIN STREAM METHODS OF BENCHMARKING
The preceding chapters have very comprehensively explored many of the most popular approaches to benchmarking performance and service within the contact centre environment. There are however additional techniques, which can be applied either separately or simultaneously that allow a view to be gained of not just the absolute performance against a specified target but to understand and explain why that performance has occurred. In other words what is it that the customer, the agent or the organisation are either doing (or not doing) that is leading to that outcome.
Such a review can take into account either directly or indirectly many of the findings or measures used within the conventional benchmarks previously discussed in order to provide a triangulated or 360 degree evaluation of performance.
It is all about Inputs and Outputs
When reduced to the lowest common denominator, efforts in a contact centre usually are based upon carefully understanding and managing the inputs into the business / operation, with the purpose of maximizing the outputs. Inputs would be considered in terms of agents, their skills, training and support infrastructure, while outputs would be considered as results these achieve, such as sales, complaint resolution, upsell, cross sell, extended customer tenure, etc. Commonly both inputs and outputs can be fairly easily measured however the area that has the greatest degree of assumption and often least proven is the connection between input activity and output achieved.
For example managers should know which inputs have the greatest effect on output achievement.
The above approach brings another key measure into play beyond productivity and customer satisfaction per se; it introduces the need to be able to understand the interaction of all of these in order to benchmark overall performance and direct resources to further develop achievements and returns.
The Overarching Principle
The approach that this understanding leads to is to take an appropriate sample of real contact centre interactions (calls), review these against a carefully drawn up evaluative framework, noting both the input skills and behaviours as well as the outputs achieved. Then using sophisticated statistical techniques, predict which factors have the most significant effect on achieving success. Benchmarking in this way provides the vital understanding of current performance and can direct further developments as we will see later in this chapter.
The importance of the conversations
Most managers would not question the importance of the conversational quality that occurs between advisors and customers, however when reviewing a list of key performance measures, or indeed previously specified metrics this factor rarely features with the appropriate importance.
The conversations that take place between organisations and their customers are a major component of the customer experience and a key determinant of its quality. Once we recognise that conversation is a vital part of the customer experience, and that customer experience is a key market differentiator, we must turn out attention to measuring the quality of those conversations. In this regard, there are two distinct aspects of quality with which we must be concerned:
1. The evaluation of the experience (was it 'good' or 'bad')
2. The distinctiveness of the experience (was it 'differentiated' or 'generic')
In order to measure conversational quality it is necessary first to understand how conversations work and what constitutes a 'good' conversation.
The traditional understanding of communication as an 'accurate information transfer' is an inadequate basis for measuring human conversation, or at least only covers part of the purpose of communication.
Accurate information transfer is only one aspect of good conversation, others include:
- Encounter regulation
- Identity management
- Management of roles and relationships
- Instrumental control
- Performative exchange
- Emotional management
- Aesthetic management
- Meta-communicative regulation
- Management of conversational norms
A good conversation is one in which all of these ten functions are handled well. Each is an important precursor or contributor to establishing, maintaining and developing the basis for co-ordinated action; our definition of good communication.
Now that we've established an understanding of the complexity of conversation, it's time to recognise that people are very good at it, and that it is a skill that can be developed and refined. But how?
Considerable research over the last fifty years has attempted to answer this question. Studies of communication competence, social skills and communication style have analysed the components required to create high quality conversations.
We have distilled a list of key conversational skills:
- Speaking activity
- Listening activity
- Interpretive understanding
- Adaptive control
- Task competence
- Information competence
- Communication confidence
- Emotional competence
- Personal involvement
- Politeness
- Rewardingness
Measuring these skills
All of these conversational skills can be deployed and measured within contact centres. This process, which involves the careful selection of several scales to monitor each skill, the training of evaluators to use those scales and the standardisation of ratings against some criteria conversations, allows the measurement of conversational quality and provides a way to quantify the customer experience.
Sampling: sizing and selecting
An accurate measure of conversational quality allows the contact centre to evaluate its current conversational performance, understand the effect this is having on business and identify ways in which conversation can be improved in order to enhance business performance.
To be useful, the measuring process must be robust and trustworthy. It is essential, therefore, that data is captured using methods that are unobtrusive and non-reactive, in order to protect against contamination. The conversations selected for analysis must be representative of all the calls taking place in the centre and must be captured discretely. If an agent knows their calls are being recorded for a specific purpose they may, for a short period of time, change their behaviours, making their calls non-representative.
Systematic and structured sampling theory must be applied in order to determine the appropriate sample size. Many quality analysis systems either underestimate the size of the sample needed, leading to error margins of plus or minus fifty percent or more or overestimate the sample size.
It is vital, of course, to select a sample size sufficiently large to be confident that the results thrown up are not born out of co-incidence or chance. Thankfully, there is a reliable reference table; a sample size of 400 will deliver results that are accurate to within plus or minus 5%.
While increasing the sample size beyond this point will deliver greater accuracy, a law of diminishing returns comes in to play. Statistical sampling tables are available to demonstrate the accuracy rating of different sample sizes.
Analysis: making sense of the data
In a typical analysis of call quality it is not unusual to produce more than 50,000 items of data. Such large quantities of data call for powerful statistical analysis and interpretation techniques that will identify the differences, similarities, relationships and patterns that exist within them.
Deriving real value from an analysis of call quality requires more than just sophisticated statistical techniques. The information these provide will only become truly meaningful when interpreted by a team of people who are able to apply sound business knowledge and an understanding of contact centre operations. It is their expertise, judgement and experience that will allow the practical implications and management possibilities arising from the analysis to be identified.
Nor is it sufficient to observe calls in isolation. The effect of the call, both in the medium and long term, must also be measured by looking at the customer's behaviour after the call. Did they use the service or become a customer? How often did they reuse the service or remain a customer, and so on.
Only by measuring post call activity, and evaluating that activity in the light of the customer experience delivered during the call, is it possible to identify the conversational behaviours that result in improved business results.
Reaping the benefits
Different behaviours will create different effects in different organisations, depending upon the kind of calls (inbound vs outbound) and the specifics of each interaction. However, it is possible in every case to identify the three or four behaviours that deliver the most beneficial effect. This allows managers to focus time and effort on the cultivation of those behaviours that matter most.
Creating a programme to monitor and improve conversational quality in the contact centre cuts across many functions within an organisation, including sales, operations, quality management, HR and customer service. In order for the programme to succeed the entire business needs to be aware of it, support it and be prepared to act upon the findings
Buy-in must begin with the contact centre agents themselves; after all, they are the people whose behaviours will need to change as a result of the programmes findings. However, it must not end with the agent. They will only be able or willing to change their behaviour if they have support and guidance from above in the form of training, coaching and even revised incentive programmes.
Conclusion
Over time managers will recognise that measuring and managing the quality of the customer experience delivered via contact centre conversations is not only possible but imperative to their success. Having embraced the importance of consistent visual and textual brand values for their organisations, they must move on to ensure that these are carried through into the essentially oral contact centre environment, which is the primary customer touch point for so many.
The emergence of robust and reliable research methodologies, will allow them to regulate conversational quality and create interactions that are both high in quality and differentiated in character, despite the inconsistencies of human nature.
In doing so they will proactively drive the creation of a more valued customer experience.