Webio
Nominated Award: Best Use of AI in Sector
Website of Company: www.webio.com
Who We Are Webio is a conversational middleware company that specialises in delivering automated and human-assisted customer conversations in credit and collections. Webio’s AI-driven conversational customer engagement platform allows organisations, who are traditionally phone, letter and email focused to have highly personalised messaging conversations in customer-preferred channels – SMS, WhatsApp etc.
Using the power of automated chatbots, machine learning, NLP, and Webio’s AI Propensity Studio, we orchestrate the conversation elements that are needed for credit/collections teams to have effective and efficient conversations that deliver optimal outcomes.
The Challenge
Consumers have many relationships with utility companies, banks, insurance companies, buy now pay later (BNPL) providers. Sometimes, a direct debit bounce or a payment can’t be made with every one of these missed payment events treated in the same way. They’re put in the process. Collections teams managing these processes have disparate legacy software, no central data-spine, and no digital communications capabilities. Engaging with customers and managing the nuances of individual customers’ circumstances especially for digital conversations at scale is challenging.
What inspired us
With a mission of rebalancing the Credit, Collections and Payment eco-system, Webio is reimaging how lenders engage and collect payments. Taking a ‘real-human view of engagement instead of hammering customers with calls to chase payments and engaging authoritatively. Meant there had to be a better way.
Webio leverages our Propensity Studio engine to predict the financial and personal well-being of every customer at each stage of their conversation and design personalised journeys, delivering the right message and level of agent resource at the right time.
Propensity Studio produces a real-time measure of how likely a conversation will end with a particular outcome, and whether that customer has an increasing or decreasing propensity to reach the given outcome – i.e., propensity-to-pay, propensity-to-be-satisfied, or propensity-to-be-vulnerable. And is specifically designed to reflect the industries workflows, its use of language, and the types of decisions that need to be made are specialised to the credit and collections contexts.
It is powerful and delivers benefits in terms of:
• Connecting with Customers–Moving engagement rates from 7% to over 35% for some clients and a 50% engagement rate where other channels have failed to deliver any engagement at all.
• Achieving Efficiencies and Increased Payments with AI – adding a layer of AI where Chatbots fully or partially automate conversations with seamless transfers to agents when needed
• Increasing Customer Engagement at Lower Costs – automating large sections of collections messaging conversations (circa 80%)
For customers, they no longer receive chasing calls that cause anxiety and lead to problem avoidance and cascading deteriorating credit situations. In messengers, customers feel they’ve got a safe place where they’ve control over the process, where it’s safe to ask questions. It changes the dynamic of interaction.
Reason for Nomination:
Webio is making a difference by giving the Credit & Collections sector the ability to understand what is happening in their customers’ conversations. Used in combination with other digital interaction technologies and easy-to-use APIs, offers a unique opportunity to remove friction from a process, enable more efficient ways to engage and remove customer distress.
Webio has a focused approach to its AI development. It’s developed as a strategic capability, as a platform, and with specific features and performances in mind. It isn’t a feature, it’s a capability.
• Webio has taken a privacy-first, security-first approach to its conversational AI. Each customer has their own instance, train their own data, and can manage the data from end to end with confidence. No black boxes, no hidden processes.
• Webio controls all the data around financial conversations from the endpoint in. Webio handles how the conversational structure is defined (Intents), the conversational workflow rules (conditions), and the conversational intelligence with its proprietary software.
• Webioflow NLP is designed for the specific requirements of the credit and collections industry. Understanding the specific terms used, the companies product names, Webioflow has been built from the ground up to ensure that we recognise entities and intents accurately.
• Propensity Studio is where credit and collections departments can use out-of-the-box propensities with no new integrations required and can re-define as needed. These propensities include Propensity-to-Pay (is this conversation moving towards a payment outcome), Propensity-to-Satisfy (is this customer likely to be satisfied at the end of this interaction), and Propensity-for-Vulnerability (is this a vulnerable customer or showing characteristics of vulnerability). Each of these propensities is a live measure available for use across the platform. Thus, an agent can see within a conversation flow, that a customer has moved from green to amber, indicating that they are now less likely to progress towards an action that will result in a payment.
• Being able to predict the outcome of a conversation might allow changes to be made in real-time that will impact the predicted outcome. Interpreting customer’s language, response behaviours and previous conversational history assists in determining how best to manage each interaction.
• In an automated conversational flow, the use of conversation propensity as part of conversational workflow rules is a deeper level of automation. Where a propensity-to-pay is below a certain number, but above another, it might be worth intervening with an agent, prioritising that customer in a different queue, or presenting the customer with another option to increase their propensity.
Webio has helped clients e.g. Oplo achieve significant improvements, some of the metrics delivered were:
– Discovered that Agent tagging in the previous system was significantly miscategorised.
– Worked with Oplo to improve conversational tagging to above 80% accuracy out of the box.
– Automation success with 80% customer engagement via Webio chatbots
– 70% response rate on proactive customer outreach
– 90% reduction in Agent Handling Times
Technical challenges
• We had to build our own NLP engine so that it could be more deeply embedded into our platform and so that it could be more easily adopted by our users.
• The NLP had to be able to identify multiple intents in the one conversational turn to reduce the number of conversational turns required to achieve a goal.
• Propensity Studio had to be developed so that users would have predictive services out of the box (again, low adoption effort) while also being able to customise and save their models. This requires full version management, tracking performance from model to model, and doing this in a manner that scales.
• The infrastructure had to be developed to support the concept of live propensities but in a manner that did not also cause costs to spiral and for the solution to be uneconomical commercially.
• The propensity model outcomes needed to be accepted and understood by the users in a manner that gave them confidence that they were accurate, but also different companies might want to use different labels, different nomenclatures, or other UX customisations.
From the Customers Viewpoint:
• Customers can self-serve their loans, arrears, and related issues entirely through conversational chat and messaging. They don’t have to jump to FAQs, Log into another portal, or jump on a phone call. It is thus far more convenient.
• Customers can engage with a financial service provider in an asynchronous conversational manner, thus reducing the level of pressure and stress a customer feels compared to speaking live with an agent on the
Additional Information:
We strategically built our Conversational AI.
•First, we built Propensity Studio that delivers predictable conversations, a unique product feature in our industry that differentiates us from the competition.
• We then focused on Propensity-to -be-Vulnerable, a particular type of measure that was not only desired by the target customers but which falls under a duty of care mandated by a regulator.
• We reused the work undertaken in Propensity Studio to deliver Webioflow, our own NLP engine, to re-enforce our position as a secure enterprise-scale vendor for the financial services sector. Again, a key differentiator in our target market.
• We then made both Propensities and our Webioflow available to the rest of the Webio platform such as workflow rules and reporting. This means customers can sort conversations by “propensity”, find out how much money is at risk, or another creative use case. Webio built our Conversational AI in partnership with our target market.
• Rather than build an AI capability in the lab we spoke with target customers and released several MVPs to assess the reaction, the usefulness of the feature and identify barriers to adoption.
• This created a new focus on ease of use. For many of our mid-tier customers to use this, it had to be a no-code interface with the minimal technical support required. For instance, an embedded NLP solution that did not require new sourcing and sign off and set up work, was an outcome from these consultations.
• Working with key clients we developed an “out-of-the-box ready” set of propensities. This required working with clients on over 100,000 separate conversations to bring the AI to an acceptable accuracy level. Webio has built a scalable data infrastructure to support a global rollout.
• During the MVP and trials processes, it became obvious that some changes had to be made to the baseline infrastructure if real-time readings were to be made available in the manner that was envisaged. This has now been put in place.
• Webio serves our customers with a Conversational AI-powered solution in the UK, Europe and the USA, enabling each to remain private, secure, and compliant within their jurisdiction.
•Webio has already sold this capability to a pan European partner who will embed this into their core offering and enable us to reach over fifty European banking and financial service
partners. Yes, yes we can.
Webio is working with leading companies today to help identify where their customers are having difficulties, where they are getting frustrated, and where they are becoming vulnerable. We believe that any company can add this digital capability to their service and change credit conversations into highly automated customer service conversations.