Webchat: The Rising Star of Digital Debt Collection
BOTS & AGENTS / 31 Jul 2023
Mark Oppermann | Chief Revenue & Marketing Director & Andy Turner | Business Development Executive
This blog is a summary of the Webio webinar: "How Webchat for Debt Collection Reduces Calls and Improves Collections"
Although often overlooked in the credit and collections industry in the past, webchat is now emerging as a powerful tool for customer engagement. With automation and AI infused into webchat, this channel has experienced a revitalisation, but the collections industry has been slow to embrace it. We discuss why this is and why Webio’s DebtChat is a leap forward for this technology.
Webchat is Not a New Concept
Webchat as a concept dates back to 1980, when the CompuServe CB Simulator became the first available online chat service. Despite its long-standing presence, webchat didn't gain significant attention until recently. Now, however, it is steadily growing in popularity across various industries, including the credit and collections sector.
Customer Satisfaction Ratings for Digital Communication Channels
One of the key reasons why webchat scores higher in customer satisfaction compared to other channels is its speed of response. Webchat brings a high level of automation regardless of the time of day, ensuring conversations with customers start promptly. This immediate response sets it apart from other channels where customers might end up waiting for responses (especially with traditional channels like phones and emails).
The Wonder and the Risk of ChatGPT and Large Language Models
The AI-powered ChatGPT is the best demonstrator of AI and the power of Large Language Models (What is a Large Language Model?) for the non-technical public. It has opened many eyes to what’s possible with AI and people are asking how they can include it in their businesses.
With impressive stats, such as 100 million users and 1.6 billion website visits by June 2023, ChatGPT has shown its potential to reinvigorate industries across the board. These are staggering stats considering it was only launched in November 2022, however, integrating AI, specifically ChatGPT, into your business requires careful consideration - tread with caution.
While AI-powered Large Language Models are brilliant, they are not without their flaws. Hallucinations - where the model generates convincing yet entirely made-up answers due to a lack of data - pose a significant challenge, especially when ChatGPT fabricates 15%-20% of its answers. This is clearly not acceptable in a credit and collections environment where accuracy is important. For instance, if a human agent made up answers 20% of the time it would certainly be a problem!
The Solution for Businesses: Specialised Custom Language Models
To address the unique requirements of the credit and collections industry, industry-specific Custom Language Models (CLM) are built using millions of real conversations as training data. These models ensure compliance, understand customer language, and identify different customer types and outcomes. Guardrails are built in to ensure the information delivered by the AI chatbots to the customer is correct and compliant.
This is the genius behind Webio’s DebtChat which makes it a perfect fit for the credit and collections industry. It has been taught to speak the language of credit and collections.
DebtChat combines AI, Webio's specialist skills, and webchat as a channel to enable compliant and scalable conversations with customers in a channel they like to use. While webchat is highly popular in industries like retail, it has not yet been fully embraced in credit and collections due to challenges like managing complex conversations and addressing customer vulnerabilities.
Historically, webchat in the credit and collections industry fell down in four areas: it was not integrated with backend systems, it lacked personalisation, it offered limited benefits to customers and there was a high percentage of agent involvement. But DebtChat has addressed these problems.
DebtChat allows for a more personalised approach to customer interactions, addressing their needs in real-time by pulling in customer data at the right time and in the right context.
It also improves collections metrics as it offers more options and better service.
Additionally, customers' satisfaction with digital channels increases, showing an improvement of 7-12% in customer engagement and efficiency.
DebtChat from the Customer's Point of View
DebtChat from the Agent's Point of View
DebtChat Performance vs. Phone and Email
DebtChat's performance surpasses traditional communication channels, such as phone and email. It reduces agent handling time by a mind-blowing 90% and automates 45%-80% of interactions. Handling up to 40% of frequently asked questions through an FAQ bot further enhances efficiency. In means that agents don’t have to answer the same common questions over and over; rather, the chatbots can easily handle these types of queries which frees agents up for more high-value issues.
Webchat, empowered by AI and automation, is transforming thec redit and collections industry. DebtChat, as a specialised offering, presents an opportunity to engage with customers in the language of credit and collections, at scale, and provide personalised experiences.
Q & A
Q: Can you share tips on how to customise webchat interactions for different types of debt and debtor profiles.
A: The key to successful customer engagement lies in personalisation and data utilisation. To achieve this, it's essential to incorporate relevant data into conversations and configure different customer journeys. Building conversation flows in the bot building stage is easy, allowing for slight modifications to create various flows for different customer types.
There's no magic involved; it's all about using the system's capabilities correctly. Creating specific journeys is simple and configurable on the system, which enables the use of live data when possible. While API connections for live data might not be implemented immediately, it's a valuable consideration to bring in real-time information such as balances and payment dates securely.
The approach is to take gradual steps and not rush into complex solutions. By steadily building up personalised capabilities, the business can achieve specialisation and ultimately deliver the best results for both the business and its end customers.
Q: What are the scalability considerations when implementing webchat, especially as the volume of customer enquiries increases?
A: The real game-changer here is scalability through automation. But let's not dive headfirst into the deep end, aiming for 75% right away. Instead, we can strategically deploy chatbots, perhaps starting on the contact page, gradually limiting the influx of users. This allows us to focus on honing our expertise and training the agents to perfection.
As we gain confidence and proficiency, we can widen the gates and open up new possibilities. The key lies in leveraging the speed of the channel, and recognising that not everyone will immediately embrace this cutting-edge channel. However, once users experience the efficiency and convenience, they'll keep coming back for more.
It's essential to learn from the success stories of other companies, understanding how they built up their capabilities internally and integrated clever technology solutions. Finding the right balance between internal expertise and technological prowess is important. The speed of your journey should be comfortable to everyone in your business, considering staff abilities, and most importantly, ensuring the end customer receives the ultimate experience.
Q: What metrics should be used to measure the effectiveness of webchat?
A: You've got the lag and lead measures and you've got those metrics that are going to indicate better performance on those key metrics.
Quite often when we do a pilot, the question is, 'Can we increase collections by a certain percentage, normally between 5 and 15% within 3 months?' Often, it's related to an inbound use case like webchat. Businesses are inundated with inbound calls, and their call rate is higher, so the goal might be to reduce inbound calls by 20% within six weeks.
Another important metric is the first-time resolution rates, which typically indicate the number of conversations completed per day versus phone or email interactions. We also consider quality of service and PS scores.
Taking a step back, the overall time to value in conducting a pilot, investing time, resources, and some money, as well as researching metrics, is crucial.
I recall a company that attempted to develop its own webchat channel and manage everything internally. After 12 months, they realised it involved a lot of technical work, and deploying bots was not quick. It consumed a significant amount of engineering time, and the results were not optimal. However, using modern bot solutions like Webio, the overall time to value is significantly improved.
Addressing one of the recurring metrics, many companies, regardless of scale, are struggling to retain and attract contact centre staff. Call abandonment and the lack of enough agents are major issues in businesses. This is where moving to a digital channel and automating processes can make a real difference.
Speed of deployment is a genuine advantage, and it's why we advocate for such solutions. By addressing these real issues, companies can improve their operations and customer experiences.
Q: How do you ensure that the Custom Language Model aligns with the brand voice and communication style?
A: Firstly, nothing gets sent to a customer that isn't pre-approved by you. So, you have full control over everything. This is important from a compliance standpoint and also for brand alignment. Whether it's a rule-based logic step or using Natural Language Understanding to read and then send back, your customers won't see anything that wasn't pre-approved.
Brand voice ultimately comes down to culture. As a business, you'll be influenced by factors like consumer duty, fair outcomes, supporting customers, empathy, and compassion when communicating with people. Even when using Custom Language Models, you can still maintain that brand voice because these models are trained to respond the way you want them to. They enable your bots to handle more of those repetitive interactions you would have otherwise messaged back.
The key is to ensure that whatever is sent back in the bot, whether it's rule-based or using custom language, is pre-approved. This way, it will be aligned with your brand's voice and guidelines."
Let Webio help define and achieve a metric that matters to you
Take advantage of our ground-breaking AI-assisted communications to achieve better outcomes for your enterprise.
Register for a personalized demo where our team will review how Webio's intelligent customer engagement solutions can positively impact inbound and outbound customer engagement in your business.