Blog

Four Steps to Artificial Intelligence in Receivables Management

Startups focusing on Artificial Intelligence are shaking up the traditional financial services and insurance industries. In 2017 to date, investments have reached over 6 billion USD – the year with the most funding thus far.

Having started implementing and running AI-based receivables and debt-collection services in 2016, collectAI brings in a respectful track-record. Implementing AI in the receivables process takes some time and you need an experienced partner for the roll out.

If you don’t have AI yet today, be ready to start a small journey in those four steps:

  1. Visualisation
  2. Digitisation
  3. Optimisation
  4. Artificial Intelligence

1. Visualisation

Transparency is a major challenge, the visibility of the claim history, status and further information is vital. This is why we implement the tracking per claim to visualise every touchpoint on our portal  through our API and a series of web-hooks. Both the client, and the customer benefit from this and it is useful if the claim continues through to later stages of the process.

Data displayed are for instance the time and date stamp regarding the overdue payment including a dispute mechanism offer. The detailed information feeds into the collective knowledge around the behaviour of customers.

At what time, on what day through which channel is a particular group of group of customers (e.g. millennials) most likely to engage with a communication.

2. Digitisation

Receivables processes haven’t kept pace with the latest digital tools available to help customers settle invoices faster. There are two areas of digitisation – communications and payments.  

Speak to your customer at the same level: Cover all the channels where in your communications matrix with instant payment options.

a) Communications

The modern customer is agile and smartphone obsessed. In Europe, the average consumer spends three hours per day with his smartphone. Therefore, it is crucial to offer a variety of communication methods such as SMS, messaging apps, emails, voice calls etc – traditional and digital. Once the customer is reached, he should receive all the details about her or his outstanding invoices including a link to a responsive landing page. Our white label service for dunning focuses on a trust-building look and feel for the customer for best conversion rates.

b) Payments

Overdue payments today are typically settled by bank transfer. The method is low cost, but also comes with a low conversion as it is not user friendly: The customer needs to log into her or his online banking to make the payment. Modern payment methods like SEPA, Sofort, credit cards easen up the process and increase the likelihood of completing a payment digitally.

Digitising the receivables process – communication methods, landing pages and payment methods generates significant improvements in performance.

3. Optimisation

The receivables process can be optimised by focusing on content, user experience (UX), segments (by age, debt size, gender, location) and more options.

Scenarios are analysed through A/B testing. One of the benefits of a fully automated platform is being able to run more scenarios.

Running multiple scenarios and workflows was a lot of manual work for the client. We were able to automatically conduct this with continuous improvements in conversion.

4. Artificial Intelligence

With a fully trackable, digitised and optimised process it is time to introduce the AI.  Different AI ‘agents’ or variables can be tweaked for better results.

Our technology uses AI to support the tasks in receivables management. This helps to determine successful communication strategies that work for both customer and companies.

The platform learns from the customer’s behaviour, and when deployed, the AI agent can automatically change the next step.

For example, the ‘Time of Day’ agent – the AI can react to an unsuccessful communication attempt at 8.30 in the morning by switching automatically to another time of day.  

This can be realised by applying the type of machine learning called reinforcement learning as well as neural networks for predictive analytics. Currently, the process is focused on three main agents; time of day, communication channel and tone of voice. Next on the list is day of week/day of month and we plan to add more.

The AI changes static scenarios into a more flexible process finding the most effective communication approach for the customer. All activities are visible, digital, automated and summarised in a report.

AI is a journey delivering improvements along the way. The harvest of your results is considerable and worth the effort: Your company benefits from a clear cost reduction, improved cash flow, and a higher customer satisfaction.

This may be of interest to you


Steve Emecz • April 18, 2017

Wells Fargo Tests AI-Driven Chat Experiences with Facebook Messenger

Are we seeing AI reaching a tipping point in Financial Services? In my talk ‘AI In Financial Services’ at

Steve Emecz • April 25, 2017

Will banking bring back the personal touch through AI?

According to the recent Accenture Banking Technology Vision 2017 report, four out of five senior banking execs (they

Steve Emecz • May 28, 2017

AI In The Real World

This Thursday I’ll be presenting a live case study of AI being used in financial services at the AI Expo at