Date Published: December 17, 2009 - Last Updated 5 Years, 184 Days, 17 Hours, 1 Minute ago
The great Business Intelligence (BI) explosion turned abstract data into useful information for business users. Contact centers followed this trend, and today it’s hard to find one that doesn’t scoop up large amounts of performance data that it turns into metrics of various forms for analyzing call times, resolutions, and workforce performance.
However, there is a need to collect agent data at a level that’s much more granular than common BI packages gather and analyze – and then to turn that real-time data into an immediate, proactive agent tool to take complex, time-consuming, critical processes out of human hands so that the agent can focus on the customer. With such measures in place for agents, the center manager could then better use other existing higher-level dashboard-type data for productivity improvements such as call coaching and workforce alignment.
Business Intelligence Challenges
Traditional contact center BI has been very useful for understanding overall metrics and workforce management. Different technical systems for collecting agent metrics are overlaid on existing application stack components such as computer-telephony interfaces (CTIs), customer relationship management (CRM) desktop or cloud applications, and other points. The data gathered on call times, numbers of repeat calls, and other output-based statistics that support service-level agreements go to management alert dashboards or are stored in databases for later analysis to drive changes in agent behavior or center operations.
But a large – and critical – gap exists at the agent level. Data collection typically can’t see into an agent’s entire suite of applications. If it could, it would offer very fast insights on some of the seemingly unintelligible aspects of agent behavior. This shortfall can be seen in some of these areas:
Lack of insight into agent activity. Tracking average handle time (AHT) is relatively simple, from tracking call beginning and end times from the switch. But contact centers have very little visibility into specific agent activities during the course of a call. Some things to consider: Why do certain call types take three minutes? What happens during those three minutes?
What applications or pages are navigated? What information is being copy-and-pasted? What application or page is in focus for the largest percentage of time? And calls and workflows might span as many as eight to 10 applications, so capturing user metrics across these applications is important for identifying process bottlenecks.
Lack of real-time visibility into compliance violations. This includes multiple types of compliance violations: government regulations, do-not-call lists, client or partner restrictions or lack of adherence to desired business process flows.
Lack of timely identification of undesired customer activity or positive opportunity. Real-time data on agent/application interactions can provide tip-offs that a liability such as an account cancellation is in progress, or an opportunity, such as an upsell, is possible, and quickly trigger responses.
There are probably other gaps that you can recall from your own experience.
Data Collection Solution Overview
Contact center analysts hunting for better performance often see several main data strategies for analysis and improvement. One, call management and analytics, is a mature field that uses tools to judge and improve call processing by analyzing the structure of the call itself. Some common goals include improving first-call resolution (FCR), streamlining customer onboarding processes, and improving up- and cross-sell processes and results. Some current technologies include:
Speech analytics. This technology, sometimes called “audio mining,” processes customer/agent voice streams to detect key phrases, moods, and procedures. The results are compiled and scored based on different customer service or agent performance criteria and are used for agent coaching, verification of script and process compliance, and determining customer moods and responses.
Call recording. Call recording technology offers managers the ability to monitor both voice and some agent application use. It is used to return to an event later in case of customer complaints, to develop agent coaching strategies, and the like. Because of legal, regulatory, and other compliance requirements, some recordings (e.g., processing credit card transactions) must either be controlled with strict and sometimes costly security or the recording must be paused at specific times to prevent storing data like personal verification codes. Some quality management solutions can do this automatically, and tag each segment for quick access. Though without that feature, it’s hard to pick a valid sample of a recording for training, coaching, or compliance purposes.
Keystroke monitoring. This mature technology logs users’ direct application input. The result can be voluminous amounts of non-contextual agent data that is probably correlated with other sources using some kind of server-side logic engine.
Server-side events monitoring. Monitoring technologies often rely on separate hosting, administration, and data storage and analysis, powered by custom code solutions. Mismatches between original applications and business processes and new applications and requirements can put server-side solutions in a constant game of catch-up. One example is the integration challenge posed by cloud-based applications that a center’s IT department doesn’t own and which have no integration points. Additionally, server-side monitoring can only drill so deep. It may show that X percent of customers leave a call before an agent can present an upsell offer. But it may miss the real problem, which could be tedious, error-prone manual agent work-arounds such as copy-and-paste because of a bewildering set of desktop, mainframe, and even virtual applications.
“Augmenting or alternate technologies.” These can include self-service channel alternatives to direct agent interaction. Among these channels are voice response units (VRU)/interactive voice response (IVR) and Web-based applications such as account access and Frequently Asked Questions, and they replace agent call time. There are some excellent cross-channel analytics platforms that nonetheless still lack granular comparison views into how agents really work with their line-of-business systems.
Real-Time Data Support
The technologies above are often included at different levels in contact centers worldwide, and can produce many gains in efficiency and revenue. However, consider an alternative: real-time data support to agent operations and execution. This approach is based on how the agent uses and interacts with the assigned suite of desktop and other enterprise applications to handle calls and resolve problems. Data-wise, it’s possible to collect each mouse movement or click of any application running on the desktop, including Web-based and streaming, such as Citrix. This data is much more granular, and can be applied in a number of solution areas, such as:
Identifying business process bottlenecks. Newer data collection technologies have made this significantly faster and much more precise than, for example, Six Sigma analysts standing behind agents with stopwatches or regularly reviewing calls and screening recordings.
And once bottlenecks and pain points are identified, it is possible, using the same technology platform used to capture the events, to quickly construct automations to remove human interaction from the process. Examples include eliminating the cutting-and-pasting of large amounts of data, concurrently eliminating data errors, by automatically synchronizing customer data between the CRM application and any others on the desktop. And taking it a step further by allowing data entered in one application to be pushed to, and update, all other applications automatically.
Identifying compliance violations in real time, and alerting appropriate staff. The monitor-and-automate ability of the technology platform described can extend any application or groups of applications to easily generate alerts, flexibly, based on virtually any user action or specific workflow sequence. For example, if an agent issues a credit above a pre-defined limit, a manager is automatically alerted.
Identifying critical actions and triggering responses, such as workflows, based on real-time events. For example:
Compliance prevention. Real-time agent data can trigger automations to restrict and report. An above-the-limit funds transfer can trigger an immediate freeze or cancellation of the action as soon as it is detected and before it can be executed. The event is detected, and complex automations stop the transaction, report the attempt to management, and immediately restrict access. Or, in a common scenario, routine but required Federal, State, or local regulatory statements, prompts, and disclosures can automatically display at any step during a call, and assurance logic can be added to confirm that the agent has communicated (orally, by fax, by email, etc.) the information to the customer.
Call escalation. Customers are routinely identified by category and value when calling, and certain transactions can alert specialists elsewhere in an enterprise who can assist or intervene. Account cancellation is one example. If an agent is servicing a high-value client though a CRM system and starts an account cancellation process, an alert automatically starts a call or screen recording and sends an SMS alert to a call center manager.
Upsell offers. Database matching logic can quickly associate customer data with sales offers on the back end, and an automation can then trigger the offer to the agent’s desktop in a timely fashion for a rapid close.
Extend the process to virtual agents. Turning real-time data into action is not restricted to human agents. Self-service operations and applications can be extended in the same fashion, opening opportunities for sales, enhanced service and compliance, and even intervention by a human agent at any point in a customer contact.
Accessing Real-Time User Data
As stated earlier, much of a contact center’s data is limited to what is provided by the switch or IVR. Most contact centers can tell you what their average call times are but have little or no hard data to document the business process workflows that occur during the call. There is a general lack of data about the interaction between agents and the (often) many applications used by an agent to handle a call.
The good news is that such data does readily exist. Every application accessed by an agent requires some form of user interface (including a browser for cloud/SaaS applications), which in turn relies upon the underlying Windows operating system as a central connection point for the user. Every mouse click, every character entered into a data field, every check box, every interaction that an agent has with an application creates an event that is communicated between operating system and application. By accessing this flow of communication between the operating system and applications, you gain tremendous insight into the real activities of agents.
There are technologies on the market today that enable you to readily access this information and put the events in the proper context. This gives you insight into what is occurring within the applications themselves, or in workflows that span multiple applications.
The events that you monitor and act on are under your total control. For example, you might want to track every agent event and pass that information into a business intelligence system for identifying process bottlenecks. You might discover, for example, that 40 percent of your average call is spent navigating multiple applications to find the same set of customer information. In this scenario, you could then employ an automation solution that pre-navigates applications to the proper customer records in order to shorten call times.
You might decide that you want to only capture real-time events when an agent is committing a compliance violation. For example, you might set up an alert anytime an agent changes a customer credit limit beyond a pre-defined ceiling. Or, you might set up an alert when an agent closes out a call without confirmation that a government-mandated disclosure statement has been read, or even if the agent fails to add call wrap-up notes to a CRM application.
As it relates to quality management, you might leverage real-time agent activity to trigger a call recording system. In this scenario, you might trigger and tag a call recording anytime an agent opens an account cancellation screen or accesses a high-value client in a CRM system.
New technologies are adding to the diversity of data available for contact center analysis and streamlining, allowing closer focus on agent activities. But data monitoring is only a part of the story; the ability to quickly gain an understanding of pain points and then quickly flip the situation to your advantage, with quick and efficient gains in agent productivity, is the real payoff of the newer approach.
Major contact centers are using many elements of this technology to not only determine where savings can be made and revenue increase, but to take positive action in improving agent productivity and performance and increasing revenues through targeted and timely upsell performance and rapid onboarding and enhanced customer retention. And they’re using it with virtually any applications on the agent desktop, retaining their current application investment, and fully compliant with foreseeable IT expansion.