Ariel and Avia Chen, Israeli-American entrepreneurs, have been on a remarkable journey that led them from their passion for technology – which started at a very young age – to launching multiple eCommerce businesses, to the innovative fintech startup, Chargeflow.

Ariel previously served as the Head of Merchant Services and later as the Chief Operating Officer at MasterPOS. After that, he co-founded an eCommerce cosmetics business along with his brother Avia and served as the CEO of Babe Cosmetics.

Avia, the Co-founder, and CMO of Babe Cosmetics, played an essential role in building the brand and its marketing strategies. Babe Cosmetics was known for its cruelty-free beauty products and innovative subscription-based service, BabeBox.

After selling the company, the brothers founded Chargeflow, an AI/ML based chargeback automation solution.

In this exciting interview with Ariel and Avia Chen, we discuss the latest trends in fraud, chargeback management, retail, AI/ML and so much more! Get fresh perspectives that can turbocharge your business and cyber security initiatives. Join us for a great conversation!

Please share a bit about the Chargeflow story. How did the idea for Chargeflow originate?

Prior to founding Chargeflow, we ran a multi-million dollar eCommerce businesses. Although the company was thriving, it encountered a significant problem with chargebacks. This glaring issue, coupled with our expertise in eCommerce, technology, and fintech, ignited our determination to find a solution, but we couldn’t find an existing one.

In 2020, we saw a huge opportunity in helping other merchants address the chargeback issues they were facing. We sold Babe Cosmetics, allowing us to concentrate solely on developing Chargeflow. Ariel’s specialized knowledge in fintech, combined with the lessons learned from our journey with Babe Cosmetics, uniquely positioned us to tackle the problems of chargebacks for merchants.

Please explain the general problems with chargeback management for retailers:

Chargebacks as a whole are essentially using a 50-year-old system that was originally developed in the ‘70s. The issue is that not much has changed in terms of innovation. While things improved in data, speed, and signal collection capabilities, issuers, banks, and merchants still follow the same processes that have been in use for decades. It’s still a very time-consuming and resource-intensive process for merchants to collect and submit compelling evidence in order to recover illegitimate chargebacks. On average, it takes about 1.5 hours of human labor on the merchant side to fight a single chargeback and likely a similar amount of time on the bank/issuer side to review it.

As you can imagine, this work quickly adds up and starts eating away at resources and operational overhead. Many merchants simply give up on a large bulk of chargeback disputes because the cost of disputing isn’t worth it, and they simply chalk it up to the ‘cost of doing business.’ In addition, gathering enough compelling evidence data usually requires integrating with a lot of different 3rd party data sources (which can be extremely expensive).

What are the core challenges that businesses face in relation to chargebacks and fraud?

More and more people are abusing the chargeback system and as online commerce continues to surge, more and more retailers are being victimized by incidents of chargeback fraud. Merchant Fraud Journal expects that chargebacks will cost merchants over $100 billion in 2023. Data shows that as the economy slows there is an increase in fraudulent and illegitimate chargebacks.

Chargeback fraud occurs when cardholders make an online transaction with their credit card and then file a dispute with the issuing bank after receiving the purchased goods or services. Chargeflow is well-positioned to help merchants and business mitigate the problem and ensure higher business margins while reducing overhead expenses.

How have these challenges evolved over time?

As the adoption of ecommerce and CNP transactions increased over the last decade, so has the scale of the problem. Merchants who are selling online have had to face an ever-increasing volume of chargebacks as a result. This puts a strain on the merchants’  revenue margins, opex, and human resources. In addition, fraud has become more sophisticated with an increasing number of merchants turning to anti-fraud solutions; fraudsters have to stay one step ahead and have also become more sophisticated in their MO.

It’s essentially a cat-and-mouse game between these fraud prevention solutions and fraudsters. Merchants have to balance turning away good customers with overly stringent fraud prevention mechanisms and avoiding fraud-related chargebacks. To compound the matter, friendly-fraud has also been increasing as customers abuse the chargeback system to their advantage. It’s a perfect storm against merchants and without the right solutions in place it becomes really difficult to conduct business online.

What new complexities have arisen?

As mentioned previously, fraudsters have increasingly turned to incredibly sophisticated methods. They are now working cooperatively in what is known as fraud-rings, where individuals who specialize in one skill work together with other fraudsters in other competencies to create sophisticated attack vectors. Fraudsters have become very adept in covering their tracks, and it has become extremely challenging for merchants to differentiate between a truly fraudulent transaction and a friendly-fraud transaction. This is crucial for fighting illegitimate chargebacks, and the only way to accomplish this is to investigate and collect evidence for each chargeback.

How have you had to pivot your business approach, if at all?

Our goal has always been to automate a very time-consuming and inefficient process. Having run a multi-million dollar eCommerce business in the past, we know firsthand the pain points merchants face with chargebacks. We wanted to build a solution that takes away all the manual labor and guesswork out of fighting chargebacks. This has been our approach from day one, and it’s been working extremely well.

What sets Chargeflow apart from other chargeback management services?

While other solutions in the market rely on manual, time-consuming processes, Chargeflow sets itself apart with its fully automated chargeback mitigation platform. By analyzing millions of data points and utilizing Generative AI and machine learning modules, Chargeflow swiftly processes and generates customized dispute evidence for each chargeback, significantly improving the chargeback win rate for online businesses. The platform calculates the projected success of the chargeback dispute by comparing transaction data and leveraging its comprehensive data analysis capabilities.

One of the unique aspects of Chargeflow’s approach is that it only bills merchants based on successful recovery of disputes in their favor. This ensures that merchants are charged fairly and only for the value they gain from the platform’s services. Additionally, Chargeflow understands the importance of meeting customers where they already are, in as frictionless a way as possible. That’s why Chargeflow seamlessly integrates with existing ecommerce platforms, payment service providers (PSPs), and other related tools. This integration allows online businesses to leverage Chargeflow’s capabilities without disrupting their existing workflows, making the process effortless and convenient for our merchants.

Please tell us about the technology behind the product:

We built a system that relies heavily on ML/AI to automate a lot of manual processes inherent in chargebacks and transaction disputes. The system uses historical data to ‘learn’ patterns in the merchant’s transaction and chargeback profile to create AND submit the evidence needed in order to ensure a successful outcome for the merchant. Furthermore, it uses data from across the entire network of merchants using Chargeflow to train these models, which ensures a constantly improving success rate and provides other unique data insights to our customers.

Developing AI and ML algorithms can be complex and resource intensive. Can you share insights into the research and development process?

Correct, this is a very resource intensive process and we are investing heavily in it. It’s important to remember that it is an ongoing process with continual refinement and testing to see if we can outperform existing benchmarks. In general, our process follows these steps:

1. Problem Definition and Data Collection: The journey begins with a clear understanding of the problem at hand. Our team identifies objectives and goals for the algorithm based on specific business goals.

2. Data Preprocessing: Some of our data sets require extensive preprocessing to remove noise, handle missing values, and standardize formats.

3. Training Phase: This is where the algorithm learns from the data. Training involves feeding the model with the prepared dataset, iteratively adjusting internal parameters to minimize prediction errors. Our team of analysts helps to train these models and validate results.

4. Model Evaluation and Testing: After training, the model’s performance is evaluated using a separate set of data not seen during training. This ensures that the model generalizes well to new, unseen inputs. Adjustments are made to improve the model’s accuracy and robustness.

5. Refinement: Developing AI/ML algorithms is rarely a linear process. Our team often goes back and forth, refining the algorithm, adjusting parameters, and enhancing the model’s capabilities.

Other thoughts about generative AI?

There’s a business behind any research and development done in ML. We’ve seen a lot of value in the ML models and it helps us to build up the data-driven approach to tailor our solution and help businesses in any industry, at all sizes, and at scale. In our case specifically, the challenge has been representing them in front of the financial institutions who are somewhat slower to adopt these technologies.

With regards to generative AI like ChatGPT, LLMs have been around for a few years and a few companies have been using them for quite some time. However, as ChatGPT’s popularity (and breakthrough in terms of quality of results) exploded recently, we’re seeing this technology come to the forefront and there is somewhat of a race to see how to best incorporate it into business processes.

How will you continue to evolve the technologies behind your product?

Chargeflow recently announced the successful closure of an $11 million financing round led by OpenView. This investment will allow the company to further enhance its technology development efforts and strengthen its growth trajectory in the U.S. market, solidifying its position as a leader in the industry.

How will you ensure that the technology stays updated?

We will continue to invest heavily in R&D. Using new data sets to train and test our ML models to see if we can achieve even higher win rates for our merchants at very large scales.

Other innovation?

We intend to integrate even more payment processors across the globe to make using Chargeflow as simple and seamless as possible. We recently rolled out a native integration with Stripe and will shortly introduce an integration with Adyen as well, two of the biggest online payment processing companies in the world. We are also rolling new products around Alerts, which should further help merchants mitigate chargebacks before they even occur!

How do you safeguard sensitive user information?

Chargeflow emphasizes the importance of user data security by aligning with international standards and regulations like SOC2, PCI and GDPR. We constantly maintain rigorous data management standards throughout the entire organization. From implementing strong encryption measures, both during storage and transmission, to setting robust access controls and practicing data minimization, we take a lot of effort to ensure user data remains shielded. Additionally, Chargeflow engages in regular audits and continuously trains its personnel, fostering a culture where data protection is central. By implementing these best practices, we not only ensure compliance, but also instill trust among our partners and users.

How else are you building user trust?

In our industry, trust comes with results. We are able to show our merchants a positive ROI when using our platform. Even our business model is entirely built on mutual success. We only charge merchants for successful outcomes, so if we don’t manage to win a chargeback, we don’t get paid for it. It’s the best way to align our interests with the merchants’ and gain their trust.

Is there a message that you would like to share with those who are invested in (or responsible for) chargeback management?

Chargeback management is about to fundamentally change, the underlying process (which hasn’t changed much in decades) will likely remain for the next few years, but more automated processes will take over.

Your perspectives on the future of chargeback management, technology, AI, and/or business?

The industry is rapidly changing, and we’re building a network for merchants to be able to better understand risks when accepting payments from consumers, similar to what banks are doing with their KYC process. The future of chargeback management will most likely be done through automatic resolution as soon as a dispute arises through a wide and direct connection between the business and consumer. On the bank side, there are already AI/ML solutions that are in-the-build, which will help them to automate the process on their end as well. We believe it will soon be a sort of AI vs. AI, eventually bringing data and decision making to rely entirely on that technology, to streamline all processes on both sides. This is why we believe that merchants need to have  enterprise grade AI at their disposal and why we’re positioning Chargeflow to be the leading solution in that respect.

Is there anything else that you would like to share with Check Point’s executive-level audience?

We’re very excited about the future of AI/ML and its applications in the payments ecosystem and will continue to invest heavily in developing innovative technology that can further help merchants and online businesses. We’d like to thank the CyberTalk.org audience for taking the time to read about our company and the future of chargebacks.