Case Study

Simply Energy leads the way with saving customers before they churn 

Client and industry

Retail energy

Customer accounts

750,000 +

Their challenge

Finding smart and cost-effective ways to retain the right customers in a highly competitive market.


Implemented AI and machine learning platform from SmartMeasures to deliver personalised treatment plans to reduce customer churn. 


20% reduction in churn

Across a number of campaigns during SmartMeasures proof of value between Nov 2020 – Jan 2021.

Reducing customer churn has always been a major priority for energy retailer, Simply Energy. Given the fiercely competitive market in this country, all energy retailers would attest to being serious about reducing churn. What makes Simply Energy a standout, according to Andrea Bernard, General Manager for Marketing and Sales, is they have recognised that their previous strategy based on churn propensity modelling was failing short. 

Andrea explains how Simply Energy innovated their approach to customer churn reduction and what this has meant for the business as well as their customers.

Andrea Bernard

General Manager Marketing & Sales

Simply Energy has been making energy simple and affordable for Australian households and businesses since 2005. Simply Energy is proudly backed by ENGIE – one of the largest independent power producers in the world and a global leader in the transition to a zero-carbon economy. 

As General Manager for Marketing and Sales, Andrea looks after all of the customer interaction points and channels for residential and small business gas and electricity as well as new energy products like solar, batteries and EVs. Ultimately, it’s Andrea’s responsibility to ensure Simply Energy’s customers are engaged and remain loyal to the brand. She comments,

“We’ve always had a big focus on retention and churn reduction. We’ve tried many times to build propensity models for churn and with limited success. So, we knew we had a problem.”

The challenge:
How to cost-effectively reduce churn in a highly competitive market

Since the implementation of contestability in most areas of eastern Australia between 2007 and 2016, new entrants have flooded the energy retail market. There are about 70 energy retailers authorised to operate in each of the major states.  

According to the Energy Council of Australia, Australian energy retailers are churning an average of  20 – 25% of their customers annually. On average each customer is worth $1,576* pa and there are roughly 8.8 million households in Australia. The revenue walking out the door each year is around $3.5b

Simply Energy has always had a focus on customer churn reduction. Each year they would devote resources to building churn propensity models. Over time Simply Energy identified a range of churn indicators such as low NPS scores or high bills and could accurately predict the level of customer churn.

“Simply Energy has reached saturation point in at least 2 out of our 5 key markets. So, churn reduction has become really important for us as the opportunity for acquisition diminishes. To address the issue in the past, we would build churn propensity models, typically on a yearly basis, across all customers.”

Andrea observed that these propensity models had no way of identifying at-risk customers, rather they grouped whole clusters of customers and treated them as one homogeneous group, which significantly limited the effectiveness of treatment strategies to save customers.

“The propensity models we used in the past were largely an academic exercise and only indicative of a moment in time.  They had to be updated every year.”

The solution:
AI and machine learning platform from SmartMeasures

When Andrea was presented with the SmartMeasures solution she felt it may be a good fit for Simply Energy given the maturity of their propensity models. SmartMeasures machine learning technology not only identifies customers most at risk of churning, but it also segments and groups customers so, fit for purpose save treatments can be applied. The SmartMeasures technology goes beyond an academic exercise of producing volumes of information that largely sit on shelves, by enabling businesses to constructively use that information to save customers.

SmartMeasures first AI engine processes thousands of signals across various data sources within the business.  No matter how ‘messy and disconnected’ those data sources may appear to be, the AI is able sort it and build an accurate picture of an individual customer’s health and the risk of them churning.

It was not the churn prediction capabilities of the SmartMeasures that delivered the results Simply Energy wanted, but the second AI engine that groups at-risk customers and guides the development of tailored and effective treatment plans. These groups are based on different aspects about the customer’s behaviours, profitability, adoption of digital services and so on.

The grouping can also be dovetailed into existing segmentation models businesses have already identified, thereby optimising their business data, and maximising the benefits from marketing spend on such models. Moreover, both the SmartMeasures AI engines continue to learn over time opening up the opportunity for ongoing improvements in a business’s capability to save customers.

“I think that’s what has really changed with the SmartMeasures approach. We now have very tailored and hyper-personalised treatment plans that are not just segment tailored or market tailored. It’s a model that learns and gets better over time, unlike our work with propensity models.” 

The implementation:
Not all treatment plans are equal

Phase 1

4 weeks

Collect data and train the prediction AI.

Phase 2

2 weeks

Develop Treatment Plans

Phase 3

16 weeks

Contact at-risk customers with Treatment Plans and measure results against control groups.

An initial trial project was established that included 50,000 active customers. The project was implemented in three distinct phases:

  • Phase 1: Collecting data and training the AI to accurately predict customers at risk of churning. Blind test results showed AI prediction achieved 83% accuracy.
  • Phase 2: Develop a range of treatment plans to retain ‘at risk’ customers
  • Phase 3: Run treatment plans for ‘at risk’ customers with control groups to measure effectiveness. Customers were treated with four different treatment plans using email and SMS over 16 weeks.

Upon reflection on the implementation of SmartMeasures and how people felt about the solution, Andrea observed that:

“At first there was a certain amount of cynicism and doubt about the product. People also felt that ‘so what’ if you can predict customer churn if there’s nothing you can do about it. Over the lifetime of the project attitudes changed. People have been impressed by how accurate the prediction has been and have become genuinely excited try something new and innovative.”

Implementing treatment of at-risk customers was a new operational activity and cost to the business which required some re-thinking. One of the major hurdles for Simply Energy to overcome was the amount of resources required for treatment plans based on phone calls or physical mail. In many cases the costs associated with making phone calls or sending letters would eliminate any ROI obtained from preventing a customer from churning. Further efforts on designing and running treatment plans focused on SMS and email for communication.

“At Simply Energy we need treatments that can be deployed very cost effectively at scale. The margins on our products are razor thin, which leaves us little room to spend resources on niche groups of customers with high-cost touchpoints like the telephone.”

The SmartMeasures team are equipped to utilise any channel of communication for the delivery of the treatments and adapt to the needs of the customer. Another key aspect of the SmartMeasures solution is its ability to identify topics or relevant reasons for contacting customers. 

It was found that using repurposed marketing material in the communications actually resulted in increased churn. If customers are dissatisfied or are considering leaving, a generic brand message will only alienate them further.

“Another issue that needed to be addressed is that any interaction with a customer has a potential to initiate or accelerate the likelihood they’ll churn. So, when we connect with a customer the message needs to be very relevant and of interest to the customer.”

The Results:
Reduced churn by over 20%

The level of churn associated with the control groups was measured and compared to the groups of customers who were treated. The best performing treatment plans reduced churn by over 20%.


Treatment Plan

Churn reduction measure

SMS Email
TP 1: Voucher based 14% 19%
TP 2: Service message 1 22% 3%
TP 3: Service message 2 22% -10%
TP 4: Repurposed marketing message -33% -42%

Key findings

Treatment plans based on communicating with customers by SMS performed better than treatments based on emails.

Carefully chosen words aligned to brand and customer insights data were more effective than giving away vouchers.

“Instead of relying on propensity models that provided limited use we now have a way of moving forward in developing effective strategies and plans to reduce churn. The initial results have provided insights into what areas we need to focus on and are likely to drive the best ROI”

Future plans

Simply Energy is currently rolling out SmartMeasures to its entire customer base. The focus will be on delivering treatment via SMS and other digital channels, so the treatment plans can be deployed cost-effectively and at scale.

Do you need to reduce customer churn?

Find out how SmartMeasures can help you keep customers and how easy it is to get up and running.