Case Study

SmartMeasures helps ENGIE revolutionise debt management

In a ground breaking initiative, renewable energy developer and energy retailer ENGIE is leveraging AI and behavioural science for early debt prediction and intervention. Overcoming the challenges associated with traditional debt prediction and collection methods, this new strategy is reducing financial stress for both customers and ENGIE.

Traditional debt collection methods miss the chance to intervene early. By the time collectors get involved, debts have already ballooned, and bad payment habits can be challenging to break.  This whole cycle leads to more missed payments and longer delays, which hurts both the customer and the company in the long run.



Retail energy


Joel Arnold
Head of Credit Risk & Collections

The challenge:
Early intervention and engagement

Joel Arnold, the Head of Credit Risk and Collections, and his team are focused on improving ENGIE’s financial health by minimising bad debt, collecting payments from overdue invoices, and maximising the flow of cash. Early intervention to prevent debt issues benefits everyone. Customers are better supported, the company has better cash flow, and it creates a better customer experience.

Early intervention, however, is a difficult challenge, as Joel highlights, “Many customers are forced to prioritise other expenses when facing rising costs, leading to delayed payments. To address this and proactively reach out before accounts become delinquent, we sought a more sophisticated approach. While our existing dunning and collections strategies achieved results across various customer segments, we aimed to add a layer of intelligence. This is where implementing a predictive model became a natural next step.”

Joel had seen a solution from SmartMeasures being used successfully to predict and prevent customer churn however, he was initially sceptical of SmartMeasures’ ability to meet the challenge of engaging customers early in the debt cycle.

“Having seen companies pitch predictive models for debt collection for many years but never seeing one that worked, I was initially sceptical. However, I am always up for trying something different and was curious to see how SmartMeasures could go.”

As the pilot program for SmartMeasures progressed, Joel’s initial reservations dissipated.

The solution:
A data-driven and behavioural approach to debt management

The pilot program for SmartMeasures was called ‘Predict and Engage’.  By identifying customers at risk of becoming future debtors early in the debt cycle and engaging with them, the program aimed to offer personalised support, including customised payment plans, potential grant opportunities, hardship assessment and other forms of assistance, preventing further debt accumulation.

The objective of the pilot was to test the impact we could have on the early-stage debt ($100 at 60 days) as well as late-stage debt ($1,000 at 180 days), including customers who had been disconnected.

We made three attempts to achieve the desired outcomes with the debt solution. The pilot program advanced through three stages:


Stage 1

To test if SmartMeasures could predict and influence late-stage debt.

The outcome Joel observed, “The pilot program could influence engagement in older stage debt and even with some disconnected customers but the intervention was happening too late to influence cash flow. Therefore it was necessary to predict customers much earlier in their debt cycle.”

Step 2

To test if SmartMeasures could predict very early debt ($100 at 60 days).

“We found we could predict customer debt very early in the debt cycle, however it required a sophisticated hierarchy of AI models to achieve the accuracy for the different customer debt types (see callout box below). We were able to identify potential debtors very early, when they had as little as $10 owing past 15 days.”, says Joel.

Step 3

To test if SmartMeasures could influence customers who are not yet at the debt target ($100 at 60 days) as well as those who are already at the debt target?

“We found we could improve engagement and cashflow for all customers who had not yet reached the debt target. We found we could improve engagement for those who had reached the debt target, however, it did not translate into a cashflow improvement”.

Joel observes, “It’s harder to assist customers when the debt has become significant.  And in a behavioural sense, it is difficult to change the customer’s bad payment behaviour once it has become entrenched.”

SmartMeasures not only identified customers at risk of becoming debtors, but it also classified them into three category types, based on their individual circumstances and behavioural characteristics (see callout box). Crafting effective treatment messages for our different categories of customers requires a nuanced approach.

“With most of the activity in this program occurring so early in the customer debt cycle, it was able to be run in parallel with our normal collections activity”

Let’s get into the detail on …

Customer debt types


The never pays

These customers have never paid a bill. They usually avoid payment plans or payment solutions. It was really important to identify this type early to try and change their behaviour or accelerate the disconnection process.


Can’t pay & disengage

This group are facing unexpected debt. They lack prior experience managing debt and may feel overwhelmed due to unforeseen circumstances like job loss, illness, or increased living costs. This group often disengages, potentially due to feelings of shame or embarrassment related to their situation. Recognising this unique situation, they’re approached with empathy and offered tailored supportive solutions. The treatment plans are geared towards support and education.


Can’t pay, will tell

These customers are in and out of hardship and generally take their time to pay their bills. They typically know what they need to do, as they have been on payment plans before. They don’t like feeling harassed and they don’t respond well to an empathetic approach either.

Once identified, SmartMeasures was able to deliver personalised treatment plans that appealed to each customer Type. The treatment messages were designed around these personality types to resonate with each group.

Rapid test and measure process

While the pilot program proved successful, it also emphasised the need for continuous improvement and optimisation. Joel emphasises the need for continuous improvement. “Optimising the program and testing new treatments is key,” he stresses. This commitment to rapid testing and measurement ensures ongoing adaptations for maximum impact.

“Rapid test and measure methodology goes beyond our standard champion challenger practices. We used control group testing to measure the impact of different treatment plans on the different customer types.  We used control groups to measure both Engagement and Cash Flow outcomes”.

The benefits:
Early success with personalised debt treatment

During the pilot program, SmartMeasures tested 12 unique treatment plans on customers. The treatment plans were designed by a behavioural specialist. Each treatment plan was tested against control groups to measure two things:


Customer engagement

Firstly, customer engagement. Joel says, “From my experience in many years of working in debt collection I understood that connecting and engaging with customers was the first step to solving the debt problem. So we set out to measure improved customer engagement. If we can get the customer to contact us, we have a dedicated team who is proficient at helping the customer get the debt under control.

Engagement was defined as the customer making an abnormal payment, signing up for a payment plan or being identified as hardship”.


Cash flow

Secondly, we wanted to know if there was a change in cashflow—how much sooner was debt being reduced or cleared?

The pilot program yielded very promising results. “The pilot provided the facts needed to support implementing this solution longer term for customers who are not yet at the debt target ($100 at 60 days). We were also able to verify that a hierarchy of AI models could maintain accuracy over time for each of the three customer debt types,” explains Joel.


The tables below summarise the outcomes for customers in the early stages of debt versus those who have reached $100 past 60 days due.

Not yet $100 at 60 days

Customer type



Debt reduced Debt cleared
Never pays 121% 20.5% 23.9%
Don’t pay & disengage 24% 6.9% 7.1%
Don't pay but will tell 6.9% 5.5% 1.9%

Treatments are very effective for improving both engagement and cash flow.
We were able to change behaviour when getting in early.  In addition to bringing forward cash, this has the added benefit of reducing the volume of customers who will later become problematic high debtors.

At or above $100 at 60 days

Customer type



Debt reduced Debt cleared
Never pays 93% 0.5% 1.2%
Don’t pay & disengage 33% 2.0% 0.4%
Don't pay but will tell -2.4% -1.0% 0.7%

For customers already At or above $100 at 60 days, engagement was effective but did not translate into cash flow benefits.

Not everything we tested worked, but we did achieve an
ROI of $2.74 during the pilot
For every dollar spent on the pilot it returned $2.74 in cash flow


Looking forward: A brighter future for energy customers


“The SmartMeasures team worked as though they were part of the ENGIE team.  They ran the solution as a managed service, including all operational aspects and reporting of activity and outcomes that could be shared with our normal business management teams.  We were already using SmartMeasures for our Churn reduction program so we were familiar with how they would run things and were able to leave them to it”.

The ENGIE and SmartMeasures partnership signifies a significant step forward in debt management within the energy retail sector. By leveraging predictive AI and behavioural science, the initiative promises to reduce financial stress for customers, improve cash flow for ENGIE, and support our customer first approach.

Do you need to reduce customer debt?

Find out how SmartMeasures can help you identify customers before they get into debt.