Leveraging Speech Analytics to Boost Compliance and Contact Center Performance


Data from 2016 shows the average U.S. household carrying debt of more than $132,000, and with more than $16,000 of credit card debt on average. Student loan debt is also exceedingly high. All of these debts unfortunately mean some borrowers have difficulty keeping up, and that’s where debt collection centers must be involved.

This consistent rise in debt has driven a corresponding increase in debt collection calls, which places pressure on centers to maintain compliance. There are state regulations to manage as well as federal licensing laws under the Fair Credit Reporting Act (FCRA), the Fair Debt Collection Practices Act (FDCPA) and the Telephone Consumer Protection Act (TCPA).

There are thousands of Fair Debt Collection Practices Act (FDCPA) lawsuits filed every year, and consistent year-over-year increases in Fair Credit Reporting Act (FCRA) violations. FDCPA violations are enforced by the Consumer Financial Protection Bureau (CFPB), a national consumer protection agency, and are levied against debt collectors that use abusive, unfair or deceptive practices in the course of business. FCRA offers consumers protection against the misuse or misreporting of credit information, which can cause a lowered credit score, higher interest rates, and denials.  Common FCRA violations can include instances of failing to report debts that were discharged in bankruptcy, reporting an account as active when it was closed by the consumer, and reporting old debts as new. Other types of violations are those laid out by the TCPA, which is governed by the FCC and covers the approved use of automatic dialing systems, artificial or prerecorded voice messages, SMS text messages, and fax machines.

Debt collection is highly regulated, and there’s increasingly a focus on fair consumer treatment. For collection agencies to succeed, they must find a balance as they stick to the compliance rules and treat consumers equitably and also use best practices to legally collect the maximum amount of consumer debt. Doing this well requires a firm with a high-performing group of agents and a keen understanding about who to target with calls and when are the ideal times to reach various groups.

Staying in Compliance

Speech analytics allows collections centers to audit 100% of the agent-customer interactions, so they can quickly spot non-compliance. This contrasts sharply to the traditional approach of performing compliance monitoring through random call sampling. Random chance means a compliance audit could have picked 20 non-compliant calls out of thousands of properly-conducted calls, or found the converse situation. In either case, the sample size is typically so small that it cannot be trusted.

Analytics platforms can be used to identify the proper Mini Miranda language, Right Party Contact language, and potential FDCPA violations by spotting the right or wrong phrasing from the agent. Fundamentally, the platforms can mark “things you should say” and “things you shouldn’t say.” It provides metrics about the agent’s ability to stay within the approved language guidelines, and the most advanced systems can determine empathy and other similar metrics that are gleaned from the context of the conversation. Such platforms can also immediately identify any abusive language from either side of the party. With such an advanced tool, managers can see which agents are engaging in potentially non-compliant behavior. They can then determine (based on other agent metrics) if the agent needs retraining or if the action calls for immediate termination. The key benefit of such analytics is it can proactively prevent large regulatory fines. Such platforms can also track DNC instances, wrong numbers, and other similar situations.

Violations of FDCPA, FCRA or TCPA regulations are time-consuming and costly. The manual sampling of calls does very little to mitigate non-compliant behavior and stop litigation. Using speech analytics helps identify non-compliant behaviors and diagnose whether they are specific to an agent or a pervasive problem. Speech analytics also makes it easy to make dynamic adjustments to policies as compliance regulations change. Managers can also use speech analytics data in positive ways to showcase agents that adhere to compliance policies. They can then use those agents’ conversations as training models to provide context to the trainings. An additional tactic is to introduce a “compliance leaderboard” where top performers are recognized and compensated for their adherence.

Real-time analytics provides supervisors and agents with a tool for immediate feedback. Instead of sampling calls and then scheduling time to talk with “Agent Smith”, the supervisor can simply see non-compliant behavior occurring immediately and intervene when needed to guide Smith on the right action. So for example if a new regulation is introduced, it can be added to scripting and speech analytics so that it can be tagged in a call transcript and returned in search results, and supervisors can immediately see if this new direction is being followed.Real-time analytics also provides agents with next best action guidance so that they can self-correct while on a call. They can be prompted with proper disclosure and right-party ID language as well as language for treating customers fairly.

Maximizing Performance and Payments

Analysis of every agent interaction, including chat transcripts, email, social media, and the phone can provide management with new insights to boost performance. They can identify agent’s phrasing and actions that result in either successful or unsuccessful collections and then utilize this information as a learning tool for the group.

Enabling such analysis are advanced speech analytics solutions that turn conversations (unstructured data) into searchable text (structured data). Such solutions provide a complete picture of agent and customer interactions. In a typical contact center, supervisors might listen in to one to three percent of every agent’s calls, take manual notes, and then sit down with each individual agent to discuss the results. This entire procedure is not only time consuming, but it’s also very unreliable, as the small sampling might not be reflective of the agent’s true performance. Perhaps the samples were taken when the agent had a bad day, or conversely they could be taking samples from the one day in the month where the agent exhibited quality performance. In either instance, the unreliability of the data means managers cannot uncover meaningful insights into how the agent’s performance and specific language and behavior directly influences debt settlement results.

Analytics allow the real-time scoring and monitoring of agents, so their actual behaviors can be coached and corrected. Armed with this information, supervisors can provide agents with feedback on how they can navigate customer objections and turn unsuccessful calls into successful interactions that do not upset the customer. For example, advanced speech analytics can monitor calls to identify situations where an agent is using empathetic language. Certain phrases or words closely tie to empathy, and agents that consistently use such language at appropriate times are typically better at connecting with clients on a more personal level. The analytics can also be used to judge the consumer’s “willingness and ability to pay”, so management can establish if the consumer will likely provide payment during a future interaction, or if there’s little or no chance of such an occurrence. Such detailed scoring gives managers context to build ranked call lists. So for example they could direct the “toughest” consumers to the best-performing agents in order to increase the collection rates, and give the more “willing” consumers to the newer less experienced agents.

Such data is not only useful on an agent-by-agent basis, but is vitally important when viewed in aggregate. Supervisors and managers can see how phrasing and empathy relate to other metrics such as first-call-resolution and data on how often agents were able to secure full payments. These positive-ROI agent behaviors can then be integrated into new agent onboarding materials and training methods to boost the metrics of the entire team.

Real-time speech analytics provides performance monitoring directly to agents while they are on a call. If an agent forgets to use compliance-approved language they can receive prompts reminding them what to say so they remain compliant.  Supervisors would still get involved for the more serious or repeat offenses, but by empowering agents with prompts that keep them on track, they are able to self-correct. Real-time analyticsalso provides next-best-action guidance on other desired behaviors such as showing proper empathy when a customer exhibits frustration, or offering products and services promotions when a customer talks about a competitor. Speech analytics can help upper management understand the specific language most likely to drive payments and revenue. It can provide insight into questions such as, if the agent’s method of asking for payment could be categorized as “direct” or “passive”? How quickly was a settlement of the debt broached? This type of insight can shift the content of call scripts, procedural actions taken by agents, and other aspects of the business.

Analytics directly increases the efficiency of calls, so agents can make more calls per hour that are targeted to the right consumers. Speed is also gained through the feedback, where the agents receive instant guidance, and then correct their behavior faster, which creates a circular loop of improvement and efficiency. This efficiency pays off on the bottom line, both through a higher amount of debt collected per agent annually, and an increase in the amount of business the agency can win from creditors.

Analytics can also uncover certain call behaviors or tactics that could be “hiding in plain sight.” Most contact centers will have some agents that are higher performers than others when it comes to the time it takes to resolve debts (recovery times) and their recovery rates. Without speech analytics data, these high performers might still receive congratulations and perhaps monetary bonuses, but the management team would not have much insight into “why” the agents aresuccessful. With text transcripts of all phone calls and advanced analytics, managers can see conversational trends among the top performers. Perhaps they are using certain phrasing while keeping the average call length under a certain timeframe. The analytics could uncover hidden characteristics that could then be duplicated with the other agents on the team through training. 

Moving Forward

Speech analytics that categorizes conversations into searchable text provide contact centers with a tool to properly balance both compliance and performance. It can spot agents who have overly long calls and need training on redirecting the calls to expedite payment or settlement resolutions. It also can encourage empathy and compassion while tracking how often agents used compliance-approved language. The results are contact centers that are less likely to receive fines and can attract more financial services clients due to their improved collection metrics. And while the customer conversation on such debt collection calls may continue to be difficult, analytics can at least lead to improvements in the customer experience, which has additional benefits of increasing collection rates and reducing agent attrition due to increased job satisfaction scoring.

By improving their business processes based on speech analytics data on all calls (instead of a random sampling of 1 to 2% of calls)contact centers can confidently change their processes, promote top performers, and increase revenue.

scott-kendrickScott has 20 years’ experience in software product management, design, and marketing for everything from shrink-wrap consumer applications to enterprise cloud solutions. He introduced the tag-line “Listen to your customers – Improve your business” which is both something he passionately embraces in marketing and production management, and how he articulates the benefit CallMiner Eureka brings to market. Prior to CallMiner Scott was Sr. Director of Global Product Management at MIVA (an internet search marketing company), and Director of Product Management and Marketing for the New Ventures division at Corel Corporation (makers of personal and business software CorelDraw® and WordPerfect®). Scott holds a BScE in Civil Engineering from Queen’s University (Ontario, Canada), and is certified in Pragmatic Marketing and SCRUM.


Please enter your comment!
Please enter your name here

Share post:


More like this

ViClarity Announces Several Promotions within Team of Governance, Risk and Compliance Experts

Team members to lead continued growth and enhanced client...

Compliance Talent Gap Challenges Credit Unions

Is your credit union struggling to find and retain...

Anti-Money Laundering Technology Key to AMLA Compliance

2020 was quite the year. The banking industry, which...