PDF File: 

CRM3_wp_stacy.pdf

Accenture

A new approach to computer simulation, which delivers shorter project durations, meets interaction center operations management challenges.

From a CRM perspective, the 1990s saw call centers evolve into complex operational units that capture interactions via fax, mobile, email, and Web. The evolution from call center to customer interaction center has been driven in part by CRM strategy objectives, and in part by the potential for cost-reduction-per-interaction benefits.

Organizations don't take lightly the decision to make major functional changes to their customer interaction capability. Business cases promising marketing benefits and efficiency gains underpin the large investment required to evolve from call center to interaction center. Once a decision to evolve has been taken, organizations then face the challenge of implementing and operating an increasingly complex customer interaction system.

Complex interaction centers must be managed in a manner that realizes the promised business value benefits of the business case. Traditional call center management techniques lack the sophistication to adequately meet the challenges of optimizing highly evolved interaction center operations. This article describes how a new, flexible approach to computer simulation delivers short simulation project durations, and provides a powerful tool to help meet the challenges of interaction center operations management.

This article outlines how intelligent use of simulation can deliver three distinct levels of value proposition:

• At a strategy level (validate a business case for change)
• At an implementation level (speed-up the time-to-value)
• At an operational level (reduce operating costs and improve service levels)

The Evolution to Customer Interaction Centers

CRM is about acquiring, developing, and retaining satisfied loyal customers; achieving profitable growth and creating economic value in a company's brand. To achieve the potential that CRM offers, many organizations have looked to evolve their call centers into genuine customer interaction centers that offer additional service channels via fax, mobile, email, and Web. Multiple channels expedite the flow of information between the organization and the customer, reducing barriers to communication, while offering the potential to reduce costs of interaction and heightening the ability of an organization to create brand loyalty.

The benefits of the interaction center business model are not limited to the substantial brand value enhancement sought by marketers. Many organizations have recognized the operational cost efficiencies of the multichannel interaction center. Lowering the cost of interactions is often a motivation for organizations, sending them along the continuum between face-to-face and Web-based self-service interactions. In some cases, e-commerce and Internet self-service cost efficiencies are 100 times greater than traditional call interactions.¹

Increasing Complexity

The transition to interaction centers has proven to be something of a minefield for operations managers. The increased choice of interaction channels have been accompanied by an increasing need to manage interaction centers to the level of operational efficiency that allows an organization to benefit from cost-reduction and service-enhancement potential. Many organizations have invested millions of dollars integrating complex workforce-management systems in order to optimize the operational efficiencies promised by new interaction centers. Enabling technologies such as interactive voice response, computer telephony integration, virtual network queuing, automated email response, and skills-based-priority routing have been introduced to help drive out the operational efficiencies. One consequence of the adoption of such technologies is a significant increase in operational complexity.

Recently a leading telecommunications company invested in a skills-based-priority-routing technology and a multimillion dollar workforce management system within the retail division of their organization. The complexity associated with managing their new operations was beyond the capabilities of their operational analysts. The new systems could not be configured to deliver the operational improvements targeted, resulting in a decision to turn off the skills-based-priority-routing functions. It is unlikely that the organization achieved the operational benefits that underpinned the business case to make the large capital investment.

Managers and consultants alike can struggle to provide solid operational evidence of the impact of implementing a new routing technology or interaction channel. Business cases for multimillion dollar investments are often based on static spreadsheet analysis, in turn founded on numerous simplifying assumptions. There is easily no way to capture the resonant complexity of a modern interaction center and provide management with a confident estimation of the operational impacts of a major strategic or operational change.

Managing Complexity

The problem faced by many interaction center managers is that current management techniques lack the sophistication to adequately meet the challenges of managing the increased operational complexity. Traditional tools such as empirical spreadsheet methods, Erlang C agent calculators, or optimization functions, require dangerous simplifying assumptions that compromise the accuracy of the analysis.

The scenario of calculating agent numbers for a multichannel, skills-based-priority-routing service center is a useful example. An empirical spreadsheet method, which relies on determining agent numbers by using averages to balance load/capacity ratio is hopelessly inaccurate because it ignores the effect of queuing variation. This simple spreadsheet approach to analysis always under-estimates agent requirements, resulting in poor customer service levels.

The Erlang C agent calculator was traditionally the industry standard for agent level estimation. Erlang C uses a formula that computes the probability of delay in a queue for calls, given a specific number of agents answering calls and a predicted level of inbound call traffic. The user inputs the number of call interactions, average talk times and wrap-up times for each time-block over the workday, along with a desired service level (e.g., answer 80 percent of calls within 20 seconds). The formula is iterated, increasing agent numbers until service level requirements are met. This approach is quick but is weakened by the need to simplify the scenario by ignoring blocked and abandoning customers and skills-based-routing efficiencies. Erlang C calculators also erroneously assume that all calls have equal priority. Unfortunately, these assumptions and simplification lead to overstaffing, resulting in lower than optimal agent utilization.

Many large workforce management systems now have proprietary optimization algorithms that refine the Erlang C calculator results for priority queuing and in some case skills-based routing. There are two problems with this approach. Firstly, the limiting assumptions that are made to establish the objective function compromise the accuracy of the result leading to either over estimating agent requirements and increase cost or under-estimate agent requirements leading to high customer queue times. The second problem is the large sample space that needs to be computed can require 90-hour plus run-times on mainframe computers, making optimization an impractical solution.

Simulation offers a quick and accurate solution to this scenario problem. A simulation approach can capture the necessary interactions, variations, and process complexity to accurately determine agent requirements by interaction type. Simulation scenario analysis also captures the economies of scale gained by multiskilled agents, and agent utilization by interaction type based on routing rules and priorities.

Computer Simulation

The proliferation of computer technology has allowed computerized simulation models to become available to decision makers in a wide range of industries. Simulation is a sophisticated and accurate means of helping managers understand the behaviors that are likely to result from decisions they make about their business systems.

A simulation model is a virtual representation of some complex business system, that exists in the form of a software program on a computer. Business process simulation was pioneered in manufacturing and supply chain analysis, and in more recent years has been applied to analyzing call center performance. The great strength of simulation models is that they show how the business system will likely behave under any given set of circumstances (a scenario).

Because a simulation predicts so accurately, it's a "virtual prototype" to test and perfect new investments or initiatives, such as call center consolidation, outsourcing, or skills-based routing, without experimenting on the company's customers.

The ability to replicate interaction center behavior accurately comes from the way simulation models capture the influence of sources of variation that will impact the center. The simulation model is therefore a considerable advance on previous methods such as static process analysis, process mapping or spreadsheet analysis. The value of understanding the behavioral impacts of variability is captured in this definition of interaction center simulation:

"A interaction center simulation model is a high-accuracy, predictive, computerized model of a firm's customer interactions. These may include traditional call center, e-channels, as well as any other "back-office" interactions such as order-to-cash or paperwork processing. The simulation helps ensure a balance of staffing, automation and business processes to optimize operational efficiency, revenue lift, and customer satisfaction."

Current Tools/Traditional Approaches

In the past, constructing a detailed call center simulation was a labor intensive exercise. Simulation project durations of two to three months were common, and the costs resulting from projects of this length were often unpalatable to prospective clients. A two to three month timeframe simply often wasn't responsive enough to make call center simulation a truly viable business decision aid for managers.

Several specialist call center simulation-modeling applications were developed to resolve this problem. These software tools proved to be useful for analysis of "old-world" call centers, but lacked the functionality to replicate the complex modern interaction centers. Additionally, the technical complexity of using simulation models makes their daily use an infeasible proposition.

The fundamental problems of long, expensive projects, insufficient functional flexibility, and technical complexity, have created barriers to the widespread adoption of simulation modeling. Consequently, the use of simulation modeling in interaction centers has been sporadic. Managers have viewed it as an asset that is "nice to have" but which does not warrant the commitment of time and expense to gain the business value.

The onus is on the simulation modeling industry to devise a solution which provides better business value to the end user. These improvements must be achieved without compromising the accuracy or predictive power of a interaction center simulation model as a managerial decision aid. If the simulation industry can provide business managers with such a tool, the time-to-business-value of strategic, implementation, and operational initiatives, will be reduced.

CICSim: The Value Proposition

The major barriers to widespread adoption of interaction center simulation modeling have been removed by the development of a new approach to simulation. This new approach comes in the form of a software tool called CICSim. CICSim consists of a user-interface linked to a simulation engine, and turns the technically complex and time consuming task of coding a simulation model into little more than a data entry exercise.

CICSim represents a paradigm shift in the way simulation modeling of a interaction center is approached. No longer will there be a:

• Need for 2 - 3 month project durations
• Need for total reliance on technical simulation specialists
• Limitation on replicating the technical functionality of the modern interaction center
• Barrier to in-house interaction center analysts and managers using simulation modeling as part of their daily decision process

CICSim represents a new standard in operational analysis. Interaction center simulation analysis can now help an organization more quickly realize the value of adopting a new technology or operating structure, as illustrated in figure 1. Simulation can assist managers to avoid time-hungry configuration and implementation problems for a new operating structure and can become part of the daily operational analysis conducted by interaction center managers.

The flexible new approach to simulation is designed to allow easy assessment of agent requirements, quickly helping management identify the required agent skill mix. Helping identify optimal agent requirements in-turn helps reduce operating costs by avoiding low agent utilization and ensures that customer service level targets are achieved. Some other applications of CICSim include:

• Forecasting service levels in processing centers with a backlog of work-in-progress, and seasonal arrival volumes
• Establishing benefits of load balancing across different centers
• Quantifying the impact on agent requirements of differentiated levels of customer service
• Investigating the rollout and impact on service levels of moving work off-shore

The following sections present two recent case studies of CICSim application.

Case Study — Contact Center Consolidations

A large Asian Telecommunications organization was negotiating with Accenture over an arrangement that would make significant changes to the structure of their call center operations. The arrangement called for contact center transformation to generate efficiency gains within existing call center operations.

CICSim was used to capture operational data from one of the Telco's call centers. This data was used to generate a simulation model of call centers that serviced calls from long-distance customers. This "baseline" simulation model was validated, and used as a basis for experimentation. What-if scenario questions were asked about the impact of proposed functional changes to the center.

The CICSim simulation asked and answered what-if questions about the introduction of IVR technology, reduction of after-call activity, and technology enabled reduction of average handle time. The scenario analyses provided accurate quantitative information about the impact of the proposed changes on center operational efficiency.

The changes were reported in terms of their impact on services levels, agent occupancy levels and potential cost savings. The analysis supported a refinement of the business case, helping the client benefit from a more optimal CRM architecture.

Case Study — Outsourcing Processing Work Offshore

A large global financial services company was in the midst of establishing a rollout plan to outsource work to an offshore processing center. The outsourcing business case planned for offshore agents to be trained using actual processing work items. The work of the offshore agents was to be quality checked by host agents, until such a time that they were considered fully competent and able to replace the host agents. A small test pilot program was conducted, and key data was captured on the following variables:

• The time-to-competency of offshore processing agents; the time-to-competency was a measure of the predicted processing time of offshore agents compared to expert host agents
• Time available for processing while on training
• Amount of resultant quality checking handled by host agents

CICSim was first used to develop a baseline model (establish the forecasted service level) of the existing host operation. Then, using the insight from the pilot program, several different rollout schedules and training scenarios were simulated. The key question in each simulation scenario was; at what point could the host agents be removed without negatively affecting customer service levels, as established by the baseline? The findings indicated:

• Targeted removal of host agents (as specified by the business-case) was too optimistic and would need to be delayed.
• Additional host agents would need to be allocated to quality checking processes if service levels were to be maintained as forecasted by the baseline. The findings were very valuable to the client organization, as a degradation of service levels would have been unacceptable in their highly competitive industry. Maintaining service levels was seen as an essential element of obtaining and retaining customers.

CICSim Methodology

Experienced simulation modelers know that it is important to follow a proven project methodology to deliver useful models. The first step in the methodology, for existing interactions centers shown in Figure 2, is to collect historical operational data from the "as is" interaction center process such as arrival volumes, handling times, and agent rosters. Equally important is capturing configuration parameters such as trunk and IVR capacities, routing network logic, and sources of variation including adherence to schedule and forecasting uncertainty. It is important to insure the integrity of the data in this first step of the methodology.

The "as is" model becomes the baseline against which all other "to be" scenarios are directly compared. Once the baseline model is created, it is simulated, and key operational metrics such as service levels, agent utilization, and customer abandonments are validated against the actual metrics reported for the historical period. This direct comparison allows the accuracy of the simulated baseline model to be established, and understood. When analyzing a greenfield business model, only forecasted data is available. Actual metrics do not exist so close collaboration with industry experts is essential. Benchmarking data and industry case studies can also assist the validation process.

The next series of steps namely: identifying the scenario question, alter input data, run simulation model, and analyze results are iterated, each time refining the scenario until the optimal solution is reported. Identifying the scenario question within the context of the new operating model is the most important step of the analysis and is where most time should be allocated. The question may simply be "what are the savings, for our customers, that multiskill base routing could deliver?" or as specific as "what is the speed-to-answer performance of a new consolidated center across a range of enquiry types?"

Often the simulation is linked to a financial model. This model maybe built or an existing model may be linked to CICSim. Simulation outputs such as agent numbers, utilizations, outsourcing transaction volumes and alike, are imported into the financial model allowing an accurate assessment of scenario costing. The simulation and financial analysis together give a complete solution to the scenario question and become the basis of a business case for change.

Summary

CRM strategy and cost reduction objectives have provided the catalyst for the evolution of interaction centers. This evolution occurred even though managers struggled to provide confident analytical support for the business cases to justify an organization's investment in the technology infrastructure required for a modern interaction center. Implementers have had to come to terms with complex new systems, and have grappled with the challenge of reducing the time-to-value of change initiatives. Similarly, operational analysts have struggled to gain the performance enhancements promised by the adoption of new channels or technology.

In the past, simulation modeling was considered as an option to help an organization overcome these problems. However, traditional approaches to simulation modeling had numerous barriers to use in interaction centers. Consequently simulation modeling has not been widely used by managers in interaction centers to support their decision-making. The simulation modeling industry has risen to the challenge, and developed new tools to overcome the barriers to use of simulation modeling. Decision makers in interaction center can now have access to more responsive and flexible operational analysis, enabling improved decision support at a strategic, implementation, and operational level. These improvements can help an organization realize its CRM objectives of attracting and retaining loyal customers.

Endnotes

1. Mark P. McDonald, Accenture Partner and MD of Accenture's Center for Process Excellence. "After the Gold Rush: New Rules for the Game", Defying the Limits, Volume 2, Page 202.