May 13, 2021
How and when should you tackle fraudulent returns?
Consumer modelling and segmentation has never been more important in retail, as the pandemic-driven ecommerce boom puts customer acquisition and retention under the spotlight like never before. With shopper behaviours under scrutiny, retailers are continuing to focus on fraud prevention and loss minimisation when it comes to returns.
Many have been inspired by ecommerce apparel pureplay ASOS, which implemented an automated system that can blacklist customers who are deemed to be taking excessive advantage of their returns policy. Some retailers are aiming to create similar algorithm-driven systems, while others are simply using their returns process to pre-empt fraudulent or excessive returning.
Preventing fraud and minimizing excessive returns is obviously a reasonable goal. However, retailers need to be sure that their efforts are well-placed and prioritized. Data from Appriss suggests the cost of returns fraud (both online and offline) in 2020 was $27 billion USD. That’s a lot of money, but against total retail sales of $4.04 trillion, just 0.6% of the value of retail sales is lost to returns fraud.
Now those are national numbers which may or may not ring true for individual businesses. However, the question should be asked, what price are retailers willing to pay in order to reduce the rate of excessive or fraudulent returns, and what capabilities do they need in order to achieve that goal?
If ASOS can do it…
Let’s cut to the chase here: The ASOS brand thrives on data. Yes, it is a retailer, but it’s important to recognise the part that cutting-edge technology plays in creating its success. They view data as a means to accelerate growth, which puts it at the centre of everything and every action taken at ASOS is underpinned by a cultural foundation of data-driven decision-making. By 2019, they had employed over 50 data scientists and machine learning engineers to build proprietary systems – bespoke to the ASOS business – that produce everything from consumer intelligence to product sales modelling and are constantly hiring more and more talent in this area. These teams are responsible for tools like image searching and the ‘cognitive analytics’, that tracks individual customer behaviour in real-time to match them with product recommendations.
The question is whether other retailers, with existing technology tools and analysis, can deliver a returns fraud prevention system that works as effectively.
The problem with data – big and small
Clearly the vast majority of ecommerce businesses simply do not have the resources to implement such an embedded programme of machine learning and purpose-built technology. Instead, they will rely on more traditional methods to assess their customer’s lifetime value and returns profile. That is absolutely fine, as long as they understand the limitations of that data.
For example, how are returns included in the assessment of customer value? Is the overall value of the customer calculated after all sales and returns have been made, and is there data showing the average impact of a return on future customer spending and frequency? Or are “serial returners” flagged as a ‘problematic’ segment on the basis of return volume or percentage alone?
The risk is that retailers miss the bigger picture. Our experience suggests that customers who make a higher number of returns often have a higher lifetime value – they shop more often and spend more with their favourite brands, and while they may be selective and have high expectations, their loyalty is deeply valuable. Customers with high levels of returns are valuable sources of information and feedback, too. Are they having to compensate for a sizing issue, or difficult to fathom photography? Understanding their returns behaviour gives a retailer clear opportunities for improvement, after which those return rates may well drop again.
In short, it’s easy to throw the baby out with the bathwater unless returns behaviour is both properly understood (i.e. what is driving the returns) and put into a longer-term context of customer lifetime value (i.e. the impact of making a return on future purchasing and frequency.) Without this data, no returns algorithm can make intelligent decisions about which customers are actually worth blacklisting.
Playing the numbers game to tackle fraud
After retailers properly contextualise and understand their returns data, they can come back to the question of fraud and excessive returning. There are a variety of non-algorithmic approaches we see used.
Free returns are a useful acquisition tool, but for customers whose spending and frequency don’t compensate for their return rate, retailers have the option of introducing charges for returns.
Careful segmentation and good data collection are essential to make this approach effective without punishing customers acting in good faith.
If returns data shows that most purchases are taking a long time to return to stock, retailers have the option of targeting their policy towards faster returns. For example, returning within a week is free, but there is a charge after the initial period. This could decrease wardrobing, cover some costs and bring stock back to sale faster, uplifting the resale potential value of returned items.
It’s still fairly common for retailers to require customers to obtain authorisation from customer support in order to process a return. This approach ensures that items are eligible for return, helping to prevent fraud, and could be used to identify problematic returns behaviours.
However, this strategy encompasses every customer who wants to make a return, making returns significantly less convenient across the board, not just for those who may abuse the system. 36% of shoppers surveyed by YouGov on behalf of Doddle in 2020 said that they would like retailers to remove these authorisation policies.
The most important calculation
Retailers have to be able to estimate both the current cost of returns fraud, the difference that any intervention would make, and most importantly the other costs of altering the returns process or blacklisting customers – reduced loyalty and potentially reduced brand equity.
If there’s still a significant gain to be made with all costs factored in, it’s well worth investing time and energy pursuing reduced fraud and returns abuse. In order to fully understand and appreciate what is driving returns and their longer term impact on customer loyalty, as well as to measure the difference such a project actually makes, retailers first need to have a digital returns journey that gives them proper visibility and the possibility of analysis.