by Robert Jew
Date Published May 22, 2012 - Last Updated May 11, 2016

Recently, there have been numerous articles, white papers, webinars, and presentations advocating for IT support organizations to make the transition from phone-based support to web-based support models (i.e., chat).

Web-based support can be broadly defined by the following three characteristics:

  1. Chat is the primary communication channel.
  2. Support sessions are initiated via a support website or portal.
  3. A remote control solution is used to diagnose and fix problems.

These discussions have generated significant interest in chat by focusing on its compelling benefits, such as improving support representatives’ productivity and reducing customer wait times. However, the adoption rate for this new support model has been very low and the success rate even lower. Most organizations are still on the sidelines, hesitating to make the transition to chat because they have not found a practical and efficient approach to successfully implementing and managing this new channel. Of those who have implemented chat, there are only a few success stories; most have failed to reap the promised benefits.

In this environment of rapidly changing support requirements and increasing customer demands, the question is not whether organizations should make the transition, but how they can make that transition successfully. When implemented perfectly, chat can deliver amazing customer experiences and significant efficiencies that will change all the rules as we know them. But to achieve that, one must first develop and execute the proper strategies and implement fundamentally different processes, tools, and metrics to drive performance improvements. A common mistake many organizations make is simply installing chat software and applying the same phone-based processes to the chat channel. Worse, some do not establish any structures or processes, hoping instead that the chat channel will manage itself.

With success hinging on implementation, organizations should apply a comprehensive methodology that addresses all the primary components of implementing a new support model, including:

  • Workforce management tailored to chat;
  • Process transformation and automation to leverage the new capabilities;
  • Reporting and analytics that drive results;
  • Quality management that impacts the bottom line;
  • Talent management, including new approaches for HR recruiting, hiring, and training;
  • Marketing and user-adoption strategies for chat; and
  • Systematic project planning and execution that brings it all together.

For the moment, let’s focus on the first component. In this article, I will introduce a new workforce management (WFM) framework that is specifically designed for managing chat, providing enough detail that IT executives and service desk managers can readily apply it. In this industry, where labor costs account for more than 60 percent of total operating costs, having an effective WFM approach is fundamental to successfully realizing productivity gains and cost savings.

When implemented properly, chat can deliver an enhanced customer experience at a much lower cost. These results are derived from two intrinsic capabilities:

  • Support representatives can handle multiple concurrent sessions; and
  • Support representatives can collaborate meaningfully and instantly.

These capabilities provide new ways for support representatives to interact with customers and with each other, but they also make chat more complex and challenging to manage than voice and email channels. Adding to the complexity is the fact that these capabilities are not equally applicable to all environments and situations. Traditional WFM processes, tools, and techniques were designed for phone interactions, which are always one-toone. Applied directly, they are simply inadequate for effectively managing and optimizing the capabilities of chat. I often ask contact center experts and service desk executives/managers how they determine the proper number of representatives to staff on the chat queues, and how they determine the right composition of resources with the right mix of skills and agent profiles. The most common response is that they do not have a structured approach. Some directly apply traditional WFM techniques, while others use trial and error or other nonquantitative methods. In fact, most do not even have the right input and output metrics to properly evaluate and manage performance. Under these conditions, it is no mystery why these support organizations have not been able to reap the benefits of chat. Without the proper WFM processes, they are often simultaneously overstaffed and underutilizing their resources, which makes chat more expensive, less efficient, and unable to provide the kind of experience customers expect.

Recognizing that chat is more difficult to manage than phone, a support organization must utilize a more robust and rigorous approach. This requires the organization first to clearly define a strategy that is appropriate for its unique environment, and then to develop a corresponding execution plan with all the necessary support processes. With chat, there are four possible support scenarios:

  • Single transactions: One support representative handles one chat transaction;
  • Concurrent transactions: One support representative handles multiple chat transactions simultaneously;
  • Collaborative transactions: Multiple support representatives handle one chat transaction; and
  • Concurrent and collaborative transactions: Multiple support representatives handle multiple transactions simultaneously.

I have developed a quantitative model that systematically determines the appropriate strategy and targets for each scenario. Start by calculating the staffing requirements for the single transaction scenario, which is identical to the phone channel. This is commonly done using an Erlang C calculator with the following inputs: volume, speed of service target, and handle time. But in order to accurately model more-complex chat scenarios, handle time must be decoupled from the total elapsed time of the sessions. For example, in a support session that lasted twenty minutes, a support representative may only have spent thirteen minutes handling the transactions, with seven minutes of idle time where he could have been working on another session concurrently. Alternatively, in that same twenty-minute session, three support representatives may have worked collaboratively to resolve the issue. In this case, the actual handle time is much greater than twenty minutes, since the time of three resources was consumed. The elapsed time is automatically measured by the chat system, but the real handle time has to be calculated manually.

Once the real handle time has been determined, the preliminary number of resources that are required, as well as the number of sessions they should handle concurrently, can be calculated. (This assumes that the available [idle] time during a session can be used to work on additional sessions.) Using a standard framework and guidelines, prioritize and evaluate the following inputs:

  • Transaction characteristics:
    • Level of complexity of support (range of issues, devices, OS, apps, etc.)
    • Call duration and handle time
    • Level of engagement (percent conversational versus transactional)
    • Workflow responsibility (percent of work done by representatives versus computer or customer)
    • Level of focus (number of high-concentration versus low-concentration activities)
  • Customer characteristics and expectations:
    • Customer type (consumers versus system administrators)
    • Customer expectations (ASA and ART targets, abandonment rate, time to resolution, CSAT)
    • Customer response time (ART, percent idle time)
  • Skills requirements and availability:
    • Technical skills requirements and availability (products, processes, and systems)
    • Soft skills requirements and availability (communications, typing, multitasking, etc.)

Please refer to the table on the previous page for an example of how a quantitative model can be used to analyze call flow, workflow, customer expectations, and business needs. The target number of concurrent sessions can be adjusted up or down based on the ratings of the customer characteristics and support representative skill levels. Some of these inputs are tracked by the chat, remote control, incident management, CRM, and/or business intelligence systems, while others are manually collected by the support representatives, exit surveys, and QA sampling.

By quantitatively determining the correct concurrent strategy and targets, the support center can maximize resource utilization without negatively affecting the customer experience. In some instances, a hybrid strategy may be used, with multiple scenarios corresponding to multiple situations. As the situation changes, management should continue to use this framework to quickly make the corresponding adjustments.

Similarly, using this methodology to estimate efficiency savings versus staffing requirements will help the support organization determine the appropriate collaboration strategy and targets. The quantitative model will provide insights that will help the organization determine which issues to collaborate on, as well as how many and which resources to invite into the session. Then one can use skills-based routing to properly staff the chat queue with the optimal number and composition of support representatives.

To ensure that the strategy and processes have been executed flawlessly, the support organization must track a number of quantitative and qualitative key performance indicators (KPIs). Use the following KPIs to gain a multidimensional understanding of the channel’s performance:

  • Speed of service KPIs (ASA, ART, service level)
    • Abandonment rate
    • Staffing and utilization KPIs (scheduling accuracy, utilization and occupancy, shrinkage)
    • Utilization and relevance of the subject-matter expert resources:
      • Occupancy = Time in Session + Wrap ÷ Total Scheduled Work Hours
      • Relevance = Contribution ÷ Participation
    • Accuracy and speed of resolution:
      • Critical Errors = Number of Transactions without Critical Errors ÷ Total Transactions
      • Issue Resolution Rate/First Contact Resolution = Number of Transactions Where Issue Was Resolved ÷ Total Transactions
      • Time to Resolution = Total Time Ticket Is Open to Closed
    • Compliance with processes and policies:
      • Compliance Errors = Number of Transactions without Compliance Errors ÷ Total Transactions

    After implementation, it is important to continuously evaluate the performance of the chat channel and make improvements based on key data and analysis. By ollowing a logical, systematic methodology, you, too, can implement an effective WFM process that will help you get the most value out of your chat channel.


    Robert Jew, senior manager of business services at Bomgar, has provided business solutions to over eighty contact centers at some of the most competitive and customer-focused Global 1000 companies. He developed processes and implemented best practices and world-class standards that resulted in significant performance improvements for his clients, such as increased customer satisfaction, increased revenue, and reduced overall costs. Robert received his MBA from the UCLA Anderson School of Management and his BS in mechanical engineering from UCLA.

    HDI members, visit the webinar archive to view a recording of Robert’s November 2011 webinar, “Chat, Collaboration, and Web-Based Support.”

    Tag(s): process, practices and processes, workforce enablement, technology


    More from Robert Jew :