5 steps to build AI into your collections strategy

Where the idea of AI was previously reserved for the big screen, it’s now made a full debut into our reality.

In 2024, 76% of Financial Services firms are considering or already using AI or machine learning. The possibilities in debt collection are endless; message writing, customer engagement, even quality assurance. You might be thinking - the end results sound great, but how do you integrate technology such as AI in highly scrutinised and regulated sectors?

Featuring insights from industry experts Duhita Khadepau, Jo Mikleus, Mike Zhou and Josh Foreman, let’s take it step by step.

1. Understand the regulatory perspective

Engaging your internal Risk & Compliance team throughout is essential to ensure your technology is considering the regulatory perspective from the get-go. 

The debt collection regulatory environment is complex to say the least, particularly in the United States. There’s a lot to keep up with, but Jo Mikleus (Board member of the global RegTech Association) shared their recent report which found that 80% of global regulators are implementing innovative technology solutions. Appreciating innovation as a need-to-have, along with the benefits it holds for consumer outcomes, they’re making strong strides to progress beyond the status quo. 

It’s not a question of whether the regulatory environment has an appetite for change, the question is how use of innovation such as AI addresses their key priorities. According to Jo, top of this list ranks the transparent use of data. This includes ensuring your data processes are rock solid, protecting consumers at every step. Before you start with AI and machine learning, look into:

• What data are we using, and how was it collected? 
• How are we baking ethical principles into our data management processes? 
• Are there any potential biases in our data set? How are we addressing these? 
• How are we building transparency into our modelling processes? 

Regulators aren’t a barrier to innovation, they’re a gateway to how the industry can progress safely and fairly. If you want to get closer to how your advancements can meet the regulatory agenda, engage them. Bringing regulators on your technology journey is invaluable to generating optimal outcomes with a very key stakeholder. 

2. Evaluate your data maturity level

Before you start exploring potential applications of technology like AI, take a deep look at your data. Determining your data maturity is key to understanding what’s possible in your business. Data maturity frameworks assess multiple factors, including:

• Management: Collection, storage and lifecycle management
• Governance: Policies, standards and responsibilities
• Quality: Accuracy, reliability and correcting issues
• Analytics and reporting: Analytical capabilities and reporting tools
• Culture: Organisational culture, training and development
• Security and privacy: Data protection measures in place and compliance adherence

During Better Debt, Duhita Khadepau recommends starting here to get a 360 view on the available data to feed into AI or machine learning models. Data issues are the largest barrier to using AI for Financial Services organisations - with 38% saying that privacy, storing and sovereignty is the challenge. Taking the time to work through a maturity framework will unearth these opportunities, making your data stronger and more usable. Refining your processes also allows you to make the most out of your organisation’s unique internal data, offering further potential later down the road. Once your data has predictive power, then you can see where the technology fits in.

3. Experiment and test different technologies

Now you’ve determined your approach to gaining quality data and regulatory alignment, it’s time to explore the technology. The tactic here is to start small and work your way up. Consider the problem you’re looking to solve, and think about whether a machine learning model or AI is really necessary. If it is, then here’s how you can work out exactly what you need.

Machine learning models

What they’re good at: Tasks that involve making recommendations based on a specific set of numeric data.

Use case in debt collection: Machine learning model that gives personalised recommendations on the best time to send emails to customers, based on individual account data and previous engagement.

AI

What it’s good at: Tasks that involve replicating human intelligence and understanding context such as customer sentiment.

Use case in debt collection: Writing collections emails to customers that test different calls to action, messages and tones. Continuously improves communications based on individual performance analysis. 

Once you’ve identified the right technology it’s ideal to check its viability in a contained environment first, then taking a progressive rollout including Beta before a full launch. This allows you to fix any immediate issues, build a stronger business case for investment and optimise the technology for the best results with a larger data set.

4. Build AI into your organisational culture 

21% of Financial Services organisations are investing more in staff AI training. Why? The business case is simple. It engages your people, fuels curiosity, improves efficiency and improves the appetite for innovation within your organisation. Building AI into your strategy starts with empowering your team to use new technologies in their day-to-day. 

Practical ways to drive AI into your team and organisational culture: 

• Nominating an executive sponsor to drive the uptake of AI at a company-wide level
• Introducing common AI tools such as ChatGPT company-wide and supporting teams to find specific use cases
• Sharing AI news in a company Slack channel or newsletter to keep up with trends and emerging technology
• Holding lunch and learn AI sessions to teach team members how they can use AI in their daily tasks and answer any questions
• Finding out if there are any existing tools used by teams that are already integrating AI or exploring these features in Beta 

During Better Debt, Josh talked about encouraging company-wide usage of AI tools by sharing use cases from different teams across the business. Take InDebted’s Finance team for example, who built their own AI-bot to provide recommendations for their accounting decisions. Highlighting a tangible initiative like this gives other teams a real insight into how transformative AI can be - which makes the jump to AI collections an easier reach.

5. Assess ROI 

With AI spending in the financial sector projected to double to $97 billion in 2027, understanding ROI is essential. It’s important to note that investment in AI is for long-term value, so look at this in line with short-term costs. Get a sense of the full picture with a thorough cost-benefit analysis. Other financial services organisations that have adopted AI have experienced a range of benefits:

• Creating operational efficiencies (43%)
• Creating a competitive advantage (42%)
• Improving customer experience (27%)
• Yielding more accurate models (27%)
• Opening new business opportunities (23%)
• Reducing the total cost of ownership (14)

When it comes to specifically evaluating the impact of your AI integrated collections, tracking your metrics are key. These should be aligned to your specific innovation and could include KPIs such as customer satisfaction rates, channel engagement and more. Also remember that AI and machine learning technology has the added benefit of naturally improving over time, as it learns with new interactions and data. The most important takeaway is to set a target that once achieved, warrants further investment.

Start now

These five steps provide an actionable framework to start exploring how new technology can create specific solutions that meet the needs of your customers. With transformative potential, take your time to understand and integrate this technology effectively and safely - keeping your customer’s experience front and centre. Strive for progress over perfection - but keep your Compliance team close by!