Featurespace stands out as a testament to the power of invention, determination, and a world-class team in the ever-evolving landscape of financial technology and financial crime. From its inception at the University of Cambridge to its acquisition by Visa, Featurespace’s journey is a story of dedication and persistence – across the full range of skill sets that came together to reshape an industry.
Featurespace’s Founder, Dave Excell, reflects on the last 16 years of our journey and shares in his excitement about what the announcement of Visa ‘s acquisition means for our customers, team members, partners and business.
How invention, ingenuity, and teamwork made the world a safer place to transact.
Today’s complex security landscape demands even more protection from criminals. Technology advancement cycles have accelerated in both time & fraud losses: from months to hours to milliseconds and from millions to billions to trillions of losses. The increased sophistication of fraud and financial crime means that our interactions with the financial services system is becoming riskier. I have always believed that we should be able to trust those interactions with the financial system rather than proceeding with a high level of caution.
At Featurespace’s core lies the simple question: how can we make the world a safer place to transact?
Think of when you last flicked on a light. You probably trusted that our power grid would deliver the energy needed to light up your room without giving you an electric shock! In my mind it stands to reason that we should demand the same level of confidence from our financial service providers – fraud protection as a given.
At Featurespace, we have worked hard to enhance and protect trust in the industry we serve and ensure peace of mind for our customers. We prevent billions of dollars from ending up in the hands of criminals.
We are truly AI-native, deploy leading machine learning, and empower the best and brightest minds. This allowed us to develop real-time payment protection with algorithms that maintain high accuracy over time.
Financial crime accounts for 40% of all crime in the UK and contributes to £2.3 billion in losses annually (UK Finance, 2023). Before Featurespace, the industry protected against payments by matching against only known fraud types.
In the fast-paced world of fraud prevention, real-time payment protection needs a consumer context – separating good from bad transactions – to remain effective.
Featurespace’s Adaptive Behavioral Analytics started with a groundbreaking concept: to analyze all transactions and recognize patterns of ‘normal’ or ‘good’ payment behavior, as well as ‘bad’ behavior. At the time, this approach was revolutionary. We highlighted the significance of time series and sequential data in making decisions based on transaction data. By employing machine learning techniques, we aimed to better understand the genuine behavior of account holders. This allowed us to detect changes in behavior and address new and evolving fraud tactics.
This vision was brought to life through the combined efforts of our data science and software engineering teams. This powerful collaboration, along with our customers, lead to the creation Featurespace’s ARIC platform.
The Genesis of Featurespace
Featurespace’s reputation for innovation is based on years of leading research, pioneering technology development, and dedication to our product-market fit for the financial services sector.
Founded from the Engineering Department of Cambridge University, my co-founder Bill Fitzgerald and I spent countless hours examining many potential AI use cases. Academically, I was interested in how the power of statistics could be applied to sequential data to understand the behavior of humans. In the physical world, by modelling our posture and movement, it is possible to determine emotional characteristics such as happy versus sad or nervous versus relaxed. In digital environments, such as online computer gaming, we can also determine who is cheating and who is not.
The early years were filled with challenges. The financial crisis of 2008 created a cautious atmosphere, making it difficult to convince potential clients to adopt a new, unproven technology. Knowing the importance of our work in protecting institutions and individuals, we persevered. Betfair’s decision to invest in us was the first concrete step in turning our vision into reality.
In 2014, we found our perfect match in the payments industry. Payment data is large, real-time, and sequential. Fraud attacks in this space had dramatically increased, mass migration to digital commerce was underway, and we were witnessing both physical card and online improvements in banking security. We seized this moment!
Scaling and Growing Together
Our initial success began to attract additional investment allowing us to expand our team and bring a range of unique expertise to our table. Our next significant milestone was the partnership with TSYS in 2016, which expanded Featurespace’s reach as a leading ally in fraud detection.
In 2017, we expanded into North America opening our office in Atlanta, GA – the transaction alley of the United States. My young family and I moved away from Cambridge to build new memories in a new country. From here Featurespace’s success accelerated. We started exploring and understanding the unique fraud landscape of our latest market—and, of course, remarking that our approach to modeling genuine behavior easily transferred to new geographies with dramatically different payment eco-systems.
In 2018, the Queen’s Award for Innovation recognized our first generation of technology (ABA) as industry leading. The results were game-changing back then: ABA saw fraud losses reduced by 40% which, for a large bank, can be up to £20-30m/year. Business flow increased with 80% fewer genuine transactions blocked by fraud strategies, demonstrating the significant impact of our technology on fraud reduction.
Despite our innovative approach, the criminals’ attacks continue to evolve. This was driven by:
- Adoption of real-time payments
- Expansion of access to credit and debit cards
- Increasingly sophisticated techniques used by fraudsters, specifically scams and adversarial AI
It was clear that victims and financial institutions needed further help. As the science of machine learning advanced rapidly and the market appetite for algorithmic-based solutions matured, the hardest fraud to catch remained elusive. The next level of protection required was for low-value fraud (under £5) with very high volume, and high-value (£1000+) with low attack volume.
Once again, we invented a technology to solve the problem—a recurrent neural network (RNN) that we launched as ADBN in 2021, leveraging those seven years of experience. With ADBN’s, we have been able to increase the amount of fraud detected by 38% without changing the impact to genuine transactions. This is a steep change in performance for our customers. This paved the way for our third-generation invention – the first large transaction model, TallierLTM.
From Atlanta to Madrid, Singapore to Johannesburg, Sydney to London. From the biggest banks to the payment processors, networks, merchant acquirers and schemes behind the world’s transactions, today Featurespace’s platform is used by all sizes of Payment Service Providers, processing small and large transaction volumes.
The Right Moment to Evolve Again
Featurespace’s focus on delivering value to customers has been crucial – we are proudly “beloved” by our clients. Finding a home in Visa is a landmark event that validates the vision and hard work of everyone who has shared their knowledge and lives with Featurespace.
With Visa, we can bring the innovation, integrity and purpose of our platform to more financial institutions and payment service providers and ultimately, stop more people from becoming victims of financial crime.
As part of Visa , Featurespace is poised for even greater heights. The integration of our technology with the resources of Visa opens new avenues for invention – the opportunities are limitless.
Our journey, driven by a dedicated team who have delighted our customers, serves as an inspiration for start-ups and innovators worldwide. It’s certainly something I wish my cofounder Bill Fitzgerald could be here with us to witness. Bill passed away from cancer in 2014. I will be forever grateful to Bill for accepting my application to study at Cambridge, where the Featurespace story began and opening his network that enabled us to get started. I’m sure he would be immensely proud of everything that we have accomplished.
Of course, this isn’t the end of the story – it’s simply the start of a new chapter in making the world a safer place to transact.
Thank you to our customers and friends in the industry for their continued belief and investment. Thank you to my family, especially my wife and sons, together with our friends for their unconditional and unwavering support. And finally, thank you to my colleagues at Featurespace, past and present, who have helped us grow from a plucky start-up to an ambitious scale-up and now entering our next big chapter!
It is astonishing to see that the company that we’ve built protects millions of people, across billions of transactions, for trillions of dollars, within milli-seconds. We have delivered on our promise to make the world a safer place to transact.
The best is yet to come!
Dave Excell
Share