Transforming Tolling Customer Service with Artificial Intelligence

AI can facilitate greater understanding of customer needs and improve experience

By Nina Silguero, Toll Technology Senior Analyst | HNTB

Tolling agencies are seeking new ways to modernize their customer service operations and enhance overall efficiency to meet evolving customer expectations. Driven by real-time account management apps and personalized digital interactions, customers now expect seamless, private and highly responsive service. Artificial intelligence (AI) is emerging as a pivotal solution to meet these demands, offering tools that can transform traditional tolling customer service with data-driven, tailored experiences.

A wide range of AI tools are available and can be leveraged to improve customer experience monitoring outcomes in a variety of ways:

Machine learning, the use of computer systems that leverage data to learn and adapt without following explicit instructions, can be programmed to alert supervisory staff and management using individual and team scores and provide one on one guidance to each agent in real-time.
Computer systems can now track emotional responses from customers, allowing a more holistic view of customer sentiment and allowing agencies to intervene in a timely manner.
These systems also can monitor and score nearly 100% of customer interactions, providing agencies with valuable data that can guide decision making on how to improve customer experience.

By implementing AI technologies, tolling agencies can not only optimize monitoring and customer service operations but also stay ahead in a competitive market. The integration of AI promises to revolutionize the way tolling agencies interact with their customers, ensuring accurate, timely and satisfactory interactions.

Adopting AI in Customer Experience (CX) Measurement

Prior to the development of AI, call center customer interactions have been measured in two ways:

Customer Satisfaction (CSAT) and Net Promoter Score (NPS), which can provide a glimpse into the overall state of the service center.
Quantitative surveys (e.g. rate your overall satisfaction on a scale of 1 to 10), which provide a positive or negative takeaway from calls over a period of time.

While these measurements are essential in providing a view of customer interactions, collections of open-ended comments can deliver a more comprehensive understanding of the service a customer receives.

Using AI algorithms can provide insights into customer emotions by analyzing their words, providing a more accurate snapshot of the customer experience. These algorithms analyze the textual and vocal expressions of customers during interactions, detecting nuances in tone, sentiment and word choice. By identifying patterns and emotional cues within these interactions, AI can gauge the overall mood and satisfaction of customers more precisely. This, in turn, can provide organizations with a more holistic view of the customer experience.

Collectively, these insights can inform both short and long-term actions for meeting goals and increasing customer satisfaction. There are many touchpoints AI can have in a call center, from self-service AI
chatbots and virtual agents, agent collaboration tools and real-time or near real-time monitoring through human-machine collaboration.

The following AI technologies have become widely adopted by service centers and can be leveraged by tolling agencies to improve Customer Experience (CX) measurement.

Interactive Voice Response (IVR): Most people have interacted with this long-used AI, Interactive Voice Response (IVR) systems, while calling contact centers. These automated systems have helped to streamline call handling times and call routing using technology that routes callers and offers self-service options. This is particularly helpful and customary practice in high volume service centers.

Predictive Call Routing: As an alternative to skills-based routing (think customers selecting queues), intuitive AI can route calls based on the behavior of customers to match specifically to agents best able to assist. This technology relies on customer behavior profiles to lend AI a hand and provide understanding of the customer journey, providing a more tailored experience. Companies have begun to embrace this technology of identifying patterns in call centers to optimize customer experience. Organizations must identify metrics to measure personality metrics to agents, average handling time and subject matter expertise.

AI-Powered Virtual Assistants: AI-powered Virtual Assistants can reduce costs and enhance customer experience by handling straightforward inquiries. Using Natural Language Understanding (NLU) and accurate speech-to-text, they minimize queue transfers and improve First Contact Resolution (FCR), thereby decreasing Average Handling Time. Many callers bypass IVRs to reach a live agent. Virtual assistants like Siri or Alexa help with common tasks and identity verification before connecting to an agent.

Conversational AI (Chatbots): An estimated 85% of consumers have some desire to contact brands at some point, up from 65% in 2019. Chatbots, which are software programs that use AI to simulate human conversation, are deployed on websites, messaging apps, mobile apps and in-app chat widgets so customers can quickly engage with website content and complete self-service options. This empowers customers to solve issues without the need for interaction with the call or walk-in center. A good chat bot, with adequately developed content, can reduce call center volumes significantly. When agents don’t have to respond to simple, repetitive inquiries, they can be tasked with other complex calls or necessary back-office processing. Chatbots have also evolved as a channel for making purchases and have the potential to handle other human-agent use cases that had previously only been completed by a person.

Customer Engagement Analysis: Customer satisfaction has traditionally been measured by surveys, which is how customers perceive an interaction at a moment in time. Customer Engagement is the ongoing emotional relationship that could be considered the sum of multiple moments or a customer’s overall emotional connections, resulting from total experiences with an organization. The more engaged, the more likely customers are to feel satisfied their call purpose was met, return to the business and refer it to others. Emotional intelligence AI can track customer sentiment during a call in real-time (e.g. for legal compliance) and more commonly, in near-real time.

Utilizing AI tools for analysis of customer engagement can help provide in-depth analytics on First Contact Resolution (FCR), call handling times and measure whether customers are having a positive or negative experience.

Traditional Agents with AI Power: Some of the AI tools available to human agents have the potential for significant efficiency and compliance impacts. AI can provide data and tailored prompts to service center agents to help quickly guide decision-making to assist customers accurately and efficiently. This analysis fills in the gaps of quality monitoring by capturing long pauses, tone of voice, etc. as well as indicators of agent behavior. Leveraging this information, AI can provide feedback via pop-up messages to help guide the flow of the call. After the call, targeted coaching and training may be offered alongside feedback to promote self-learning while AI-generated scorecards can help track department performance.

This human-machine partnership may also result in reduced call times, shorter wrap-up time and more positive, personalized customer experiences. Technology also can track things like how many times a customer calls to dispute a transaction and provide a churn (customer loss) risk score so agents can utilize tactics for proper call handling and customer retention proactively.

Considerations for Technology Implementation

Procuring and implementing AI technologies for tolling agencies requires strategic planning and stakeholder coordination. When planning for the deployment of AI-powered solutions and to maximize their efficacy once implemented, it is essential to consider the following aspects:

Managing stakeholder expectations—within the tolling agency, at call centers or amongst the travelling public — is essential. Some technological benefits may be immediate; others may appear over time. Communicating clearly with all stakeholders on goals, processes and timelines can ensure that all are aligned.
Fine-tuning algorithms can take time as call volume dictates the speed of machine learning. The more calls, the more quickly that machine learning can be implemented. Plan for at least 60-90 days or more for algorithms to adjust.
Spend the time adjusting and revisiting strategy for customer service metrics. Using new data means legacy measurements and scores will need to be recalibrated and staff will need to be trained.
AI technology must work with in-house legacy systems and applications which could result in IT lift or the need for hardware/software upgrades.
Dedicating IT staff–in-house or through a third party for support during the implementation and likely afterward.
Plan for data cleaning (the process of correcting or removing incorrect, corrupted, duplicate or incomplete data).
Potential for additional computer power and data storage needs. Since AI capabilities need to be fed large amounts of data to continuously improve, evaluating capacity and processing needs is suggested.
Plan for lead time and preparation for restructuring business processes, integration, training and licensing costs.
Depending on desired functionality, some vendors provide an array of services but there are opportunities to utilize multiple third-party providers to obtain the best array of AI tools.

Agency RFP (Request for Proposal) Impacts

Once an agency has decided to procure and implement AI technologies, they will need to consider the impact on how contracts are solicited. Organizations may plan to layer the data collected from using AI tools collectively with traditional CX measurements to create the most complete picture, conduct accurate quality assurance (QA) and focus on set metrics and organizational key performance indicators (KPIs).

Each agency’s strategy should be based on what the agency is trying to accomplish and then build the structure from there. For instance, if real-time monitoring is not needed for compliance, do not include it. Agents can have “real-time” AI support, but real-time sentiment monitoring may not be necessary unless actionable. Having an accurate recap of QA for the day from all staff is a significant improvement from what is widespread practice today.

There are components that should be considered for inclusion into Request for Proposals (RFPs) for Customer Satisfaction and Customer Experience tools:

Adjust requirement percentage for analyzing calls to 99%
Adjust requirement for percentage of recorded calls to 99%
Keep requirements modern
- CSAT, Service level, QA, Abandon, FCR, Customer Complaint Volumes, Agent Satisfaction, Attrition Rate, Customer Effort, Schedule Adherence
Call transcription accuracy dependency on mono versus stereo recordings
Provide examples and sample use cases so proposers can think through and offer proof points
Consideration of mutual NDAs is frequent practice
Privacy Impacts
- Voice Signature inclusion as Personally Identifiable Information (PII)
- Customer Opt-outs — this is a manual function on telephone providers
- Opt-in/out of voice data into ChatGPT — Voice data can be stored in the cloud or deleted once transcription occurs
- Development of new contract language

By thoroughly detailing requirements and expectations, agencies can ensure they are supported by effective technologies that directly align with and support their specific goals. This approach helps in selecting the right vendor and solution, paving the way for seamless integration and optimal performance.

AI Pricing Structures

With a wide range of AI options available, pricing may vary significantly by vendor, agent totals and services selected. Many vendors offer everything from a full suite to “a la carte” options.

Pricing structures may include:

One-time start-up fee(s)
Implementation costs
Training for self-service
Optional IT Staff to manage, train and fine
tune algorithms
Monthly Costs Per Interaction
Monthly Per Agent costs
Monthly billing with potential discounts for upfront annual payments

Expertise

Featured

About HNTB

Events

Featured Insights

Latest Insights