Broadening oversight of customer interactions with tech and AI

Broadening oversight of customer interactions with tech and AI webinar recap

In recent weeks, we brought together Consumer Duty experts, in partnership with the BSA, to consider all things Consumer Duty and what it means for the mortgage market. 

In our concluding webinar of the three-part series, Garry Evans, Recordsure’s Chief Product Officer, and Olivia Fahy, Recordsure’s Senior Product Manager, considered how the sector can use RegTech to effectively support Consumer Duty compliance and provide evidence for the FCA’s “show me, don’t tell me” approach. 

Here we take a whistle-stop tour of the webinar’s highlights, including a closer look at the role that tech and AI can play in helping get the mortgage market ready for the new Duty.

Watch the webinar on-demand:  

A better method of oversight than random samples

To put it succinctly, there are two problems with random sampling – the method currently used by many mortgage Quality Assurance teams: it’s random, and it’s a sample.

By reviewing just a small sample of customer interactions, a large proportion of consumer data is left un-analysed, which can make it difficult to paint a true picture of where real issues may lie. Often, there is the myth that upscaling to review 100% of all customer calls requires a dramatic increase in resource. This scenario may certainly be true if relying on human resource alone, but utilising the right kind of intelligent technology will mitigate this need – and actually give QA teams 100% oversight of customer interactions. In turn, this means their focus can be on those calls that need most attention. Our client evidence suggests that using ConversationReviewAI doubles team productivity.

Speech analytics with a difference

Ensuring good, consistent consumer outcomes will be more important than ever under the FCA’s new Consumer Duty. And that’s where Recordsure’s ConversationReviewAI can revolutionise how QA teams achieve call quality assurance whilst gaining full oversight of customer interactions.

The ConversationReviewAI model leverages machine learning and thousands of hours of manual training by our in-house linguistic analysts. The analysts meticulously examine customer calls and teach the Recordsure AI model to understand the context around the language used in conversations. 

And Recordsure’s unique ‘neural network’ solution goes beyond speech analytics that rely on identifying key words and phrases. Such models often only achieve a true positive for every five false positives. Recordsure’s solution goes way beyond this, with accuracy around the 90% mark.

Effectively oversee 100% of customer calls

To capture your firm’s customer interactions, Recordsure can do this in one of two ways

  • An API straight through to your company’s telephony systems
  • Via an audio capture app which can be used in face-to-face or remote customer conversations

The captured audio is converted to a transcript for analysis and then visually classified by our ConversationReviewAI model, where results can be seen on the Recordsure portal. The AI breaks the call down into key topics  so it’s easy to see what has been discussed when and where. From here, QA reviewers can select from classified topics to align with their checklist and systematically jump to the relevantly themed parts of the audio. The transcript aligns to the selected section so that reviewers can read, as well as hear, their chosen topic and area.

Classifications and beyond

Our AI goes one step further than classification. ConversationReviewAI will break your audio’s classifications down into further individual risk points, so it’s even simpler for the reviewer to check associated areas with the most relevant content.

As 100% of customer calls are reviewed, our model will scale and group calls based on their level of risk, meaning reviewers can spend the majority of their time looking at the most problematic calls. This gives invaluable oversight and also flags high-risk areas whilst allowing a deep dive into specific cases.

This kind of analysis would not be possible via random sampling alone. ConversationReviewAI’s scope to find trends and themes in interactions gives the tools to address the root cause of problem areas, so continuous improvement across your customer journeys can be made.

AI and tech’s role for the mortgage sector

  • Catching the errors: With mortgage teams under increasing pressure, being able to evidence the information supplied to customers for both remediation and regulation purposes is key. Taking proactive action now to gain a deeper understanding of all the advice being given to customers will prove invaluable in the future.
  • Complying with the Consumer Duty: The step-change in regulation that the Consumer Duty will bring should not be under-estimated. More and more the FCA talks about firms embracing AI and using data to gain greater oversight of customer interactions. Having this level of oversight is no longer a pipedream and is achievable, so the FCA will be keen to see this in practice. 

Breaking down barriers

ConversationReviewAI’s sophisticated system supports natural language. The best customer conversations can’t be scripted, and our contingent model is trained at analysing just that. It’s designed specifically to analyse organic, unstructured conversation – the model’s ‘engine’ has been taught from thousands of hours of transcription and annotation.

As Recordsure is used by a number of high street banks and mortgage associations, we have created a ‘best in class’ model that the mortgage market can literally plug in and go. This means no lengthy induction process, upfront investment, or leaps of faith from firms.

Our free 30-day trial aims to show just that. If you’d like to learn more about how Recordsure’s ConversationReviewAI can help transform your QA team’s call reviews, get in touch today.


Ready to get started?

Book a demo with us to experience the power of ReviewAI in action.