Learn how Mortgage Advice Bureau adopted an AI solution to speed up its complaints process and deliver excellent customer service on AWS.

Benefits

  • 75% reduction in inital response times
  • 5 weeks to build proof of concept
  • 30-second automated ingestion of complaints details
  • Reduce complaint handling to 10 minutes

Overview

Mortgage Advice Bureau (MAB) is an established, multi-award winning, UK mortgage intermediary that provides advice, both through its own branded brokers and through other brokers. It processes about 12,000 mortgages a month and receives around 400 complaints per year. Those complaints, however, are a serious matter, because MAB is regulated by the UK’s Financial Conduct Authority (FCA), whose Financial Ombudsman is the last resort for disputes. MAB wanted to ensure that it could resolve complaints in a timely manner and before they were escalated to the Financial Ombudsman. It engaged AWS Partner BJSS to help build a solution using AI on Amazon Web Services (AWS) to speed up the processing of complaints, reducing initial response times by 75 percent.

 

“BJSS was the right team for us…I’m quite technical, and I could see BJSS knew their stuff and we got on great. They’ve proven themselves.”

 

- Lucian Morris, Chief Information Officer, Mortgage Advice Bureau

“The collaboration between BJSS, AWS, and MAB was nothing short of exemplary.”

 

- Camilla Sammut-Powell, AI Specialist, BJSS

 

Opportunity - MAB needed to improve the complaints process

MAB provides mortgage advice through its network of more than 2,000 advisers across the UK, providing both face-to-face and phone services. It handles about £16 billion of loans annually and was the first mortgage intermediary—an organization that brings together mortgage borrowers and mortgage lenders but does not actually loan funds—on the London Stock Exchange. It was listed on the Alternative Investment Market (AIM) in 2014 and is focused on providing high quality advice and services to customers.

The company advises on mortgages, explaining options to customers and helping them to determine what works best for them. It also offers guidance on suitable mortgage protection and insurance options, and helps customers handle the application process.

MAB was not overwhelmed by the volume of complaints it was receiving, but wanted to improve the way it processed them. The company wanted to resolve those complaints more quickly and ensure consistent categorization, prioritization, and initial summarization of complaints.

Being regulated by the FCA, MAB needed to meet the standards set by the regulator. This includes standards relating to customer service, including complaints management. MAB takes the matter so seriously that the Complaints Resolution team is part of the Compliance department. This is because if MAB cannot satisfy a customer complaint, the customer can escalate it to the Financial Ombudsman, which the company wanted to avoid.

When a claim was received, it went to a general inbox and would sit waiting until a complaint resolution specialist picked it up and actioned accordingly. Lucian Morris, Chief Information Officer (CIO) at MAB, thought that using AI could help speed up the process of ingesting and processing claims. “There are basically two types of complaints: ‘I received poor advice,’ or ‘You provided poor service,’” says Morris. “There isn’t a lot of variation in the basics, although each complaint is different in the specifics. When we received complaints, it took our resolution team up to four days to reply when busy, so we wanted to improve this.”

Solution - BJSS bring AI expertise to bear, delivers rapid results

MAB already had an existing relationship with AWS Partner BJSS, who Morris engaged to help rearchitect MAB’s systems when he joined the company. He turned to BJSS again for help with speeding up complaint resolutions. “BJSS was the right team for us,” says Morris. “I like to have local suppliers so that we can get together in a room with a whiteboard and hash things out. However, it’s about more than just location. I’m quite technical, and I could see BJSS knew their stuff and we got on great. They’ve proven themselves.”

 

In the spirit of the project’s goal—speeding up complaint resolutions—BJSS and MAB undertook a fast-paced development. “The proof of concept for this solution took about five weeks to build,” says Dan Winfield, marketing campaign and partner manager at BJSS. “That’s incredibly quick.”

 

The key to this efficiency was the way the teams worked together. “The collaboration between BJSS, AWS, and MAB was nothing short of exemplary,” says Camilla Sammut-Powell, AI specialist at BJSS. “Intensive immersion sessions integrated the three entities into a single, cohesive team. The blended team embarked on rapid experimentation, and after a two-day hackathon, the core elements of the solution were already in place. In a mere five weeks, the teams delivered a solution that surpassed all expectations—a true testament to the deep expertise, innovation, and passion of everyone involved.”

 

Complaints are often a long chain of nested emails documenting the customer’s attempt to resolve matters at a lower level prior to escalation. These could take time to read and understand, after which point the relevant details were extracted and entered manually into MAB’s system. Once that was completed, the agent created a reply acknowledging the complaint and proposing an initial path to resolution to the customer. This acknowledgement process could take up to four days, and MAB felt that customer expectations could be significantly improved by providing an initial response quicker.

 

The team developed a system using AWS services. The solution automatically ingests complaints from emails, extracting all required output fields using Claude 3 Sonnet, a large language model (LLM). This is provided by Amazon Bedrock, a fully managed service that offers a choice of high performing foundation models (FMs) from leading AI companies.

 

The system uses additional AWS services, including Amazon Relational Database Service (Amazon RDS), an easy-to-manage relational database service, and AWS Lambda, a serverless compute service that executes code in response to events and automatically manages compute resources. For storage, it uses Amazon Simple Storage Service (Amazon S3), an object storage service offering industry-leading scalability, data availability, security, and performance.

 

Outcome - AI helps reduce complaint handling time to 10 minutes

MAB’s AI-driven solution has significantly reduced the manual effort required for complaint processing, improving efficiency and prioritization. The company is now using AI to automate the ingestion of complaints—which happens within 30 seconds of receipt—then summarize, prioritize, and categorize the complaint, identify the customer case on its broker platform, and propose an initial response to the customer.

 

Complaints management agents have all the information they need in front of them, enabling them to focus on determining the best resolution for the complaint. Instead of waiting up to four days, customers can receive a response from a human agent as quickly as 10 minutes after submitting a complaint. Best of all, this isn’t just an acknowledgement of receipt: it’s a summary of the process that will be followed and it provides the customer with the opportunity to input additional and relevant information.

 

Such a rapid response, says Morris, puts customers at ease by showing them that MAB is serious about helping them. “People don’t go to a broker because they’re great at taking data and putting it into cells in an application,” says Morris. “Most people go to a broker because this is the single biggest purchase of their life, and they want somebody who understands where they’re at. They want a broker who understands the market and can empathize with them.”