To get us started, could you provide our readers with a quick introduction to The ai Corporation (ai)?
PB: The ai Corporation (ai) is trusted around the world for developing innovative technology that allows our customers to create predictable success and grow profitably. Founded in 1998, we have a long track record of providing fraud and payments solutions to some of the world’s largest financial/payment institutions and international merchants.
Our longstanding business partnerships are based on making things simple and explainable, both technically and commercially. By focusing this way, we constantly strive to help our customers create highly profitable returns.
EazyFuel™ is our fully managed, omni-channel, cloud-based payments platform for the fuel sector, which supports a complete range of issuing, acceptance, acquiring, switching, processing, risk and analytics functions for fuel card providers and fuel retailers. Offering ultimate agility, scalability, and configurability, EazyFuel optimizes payments processing for retailers and issuers of all sizes, from the independent dealer to the multinational.
How important is POS data in fuel retail?
PB: In any retail setting, point of sale or point of purchase (POS/POP) data is extremely valuable to any data driven business. Capturing and storing POS data in a timely manner means that it can be used to drive improvements and provide insights into numerous business critical processes, including inventory management, fighting fraud and risk protection, or understanding customer behaviour.
When a data set is scaled out over an entire network of fuel retail sites, there is no better signal of a retail customer’s needs. Data can be leveraged to better understand buying behaviour at the customer level, predicting when and where a customer will make their next purchase, and can even be used to optimise business processes, such as site and network management.
What sort of insights can be drawn from transactional data for example?
PB: On its own, transactional data should be taken at face value; for example, informing us about the purchase and the method of payment. However, the true value of this data comes from aggregating it into a customer profile.
Each transactional data point provides a new insight, which can help to build a detailed picture of individual customers. Customer profiles can be created and used to identify unexpected or abnormal transactions, which is very helpful for detecting fraud. But creating a robust customer profile also enables us to gain a rich picture of a customer’s likes, dislikes, propensity to purchase and what could improve the chances of them buying again (loyalty).
For fuel retailers being able to quickly identify changes in customer behaviour via transactional data becomes crucial in ensuring customer retention and optimising card usage.
Are those insights also available ahead of a purchase?
PB: Purchase behaviour provides the most reliable data source for building and maintaining a customer profile. However, customer onboarding and CRM systems store important data too. Aggregating those different sources of data, and then enabling the right people to access that data and draw insights from it is the key to creating a data driven operation.
Information about the fleet, such as size and vehicle type for example, are important variables that help to explain purchasing patterns. Telematics is another data source that can help to generate predictive insights. However, the purchase itself is the ‘gold standard’ of data in this sector and is the main element used to build the customer profile and therefore generate a variety of insights.
How valuable are predictive and prescriptive analytics?
PB: Predictive and prescriptive analytics are extremely valuable because they are a potential differentiator for a retailer. Effective use of both predictive and prescriptive analytics can have a positive impact on the service provided to customers. Predictive analytics can provide predictions on whether a transaction is risky or help to forecast expected purchases or purchasing patterns either for individual customers, for specific sites, or for the whole business.
That level of insight can be used to drive better decision-making. For example, whether a transaction should be declined depends on whether a transaction risk score (generated via predictive analytics) exceeds a certain threshold.
Prescriptive analytics goes beyond predictive analytics and offers a path of action that can be adopted with the goal of optimising some business outcomes. That process often uses predictive analytics but will typically also incorporate some other form of analysis, such as A/B testing or simulation modelling, to evaluate possible alternatives and find the most desirable outcome.
How do fuel cards help on that front?
PB: From a data perspective, fuel cards come with enhanced data capture, such as vehicle identification or odometer readings, which help to track and manage fleet maintenance and efficiency. They can also provide an interface for telematics data, which drastically increases the ability to prevent fraud and monitor fleet needs.
Fuel cards also offer retailers the chance to engage with their customers in novel ways. One of the biggest attractions for customers remains the discounted pricing levels available via fuel cards. However, as well as the usual benefits of fuel cards for businesses, such as the elimination of receipt and expenses claims, fuel cards are starting to offer loyalty programs for drivers and fleet managers.
Making these loyalty programs effective is a key challenge for fuel card issuers and can only be achieved with mature use of data analytics.
What role can AI and ML play in that process?
PB: AI and ML are important elements of any data driven business, but they are most effective as the final component of an overall data and analytics strategy. Ensuring that data is being captured, stored, and consumed in a meaningful way are important first steps. When those steps are in place, AI and ML offer most value when a clear business problem has been identified and articulated. Provided the question and the data are aligned, AI and ML can start to generate significant value.
AI has been central to the fight against fraud, but what opportunities are there to take AI beyond fraud prevention?
PB: There is no doubt that AI has been and continues to be central in the fight against fraud. Identifying patterns that humans might miss. Humans are still central to that process, but AI enables the scale of analysis the sector needs.
Within the fuel card sector, AI can be used to identify customers at risk of attrition, using multiple signals within the data to help to predict, in advance, when a customer will close their account. Those signals can provide vital information to account managers and enable them to pick up the phone to try and understand why that customer is behaving in a certain way.
AI can also be used for customer onboarding in the fuel card sector to help determine pricing proposals and restrictions, which can be optimised for the behaviour of the fleet. There are also opportunities in the development of recommendation engines in the fuel card sector, like the systems used by Amazon or Netflix.
How has the pandemic effected the fleet and fuel card ecosystem’s journey to digitisation?
PB: Just as in the wider retail sector, the pandemic has accelerated many digitisation journeys, putting great emphasis on effective use of data and analytics to help acquire new opportunities. The pandemic has also highlighted the importance of the B2B market in the retail fuel sector. There is growing focus on ensuring that the range of added value offerings, such as customer loyalty, apply equally to B2B, as they would in the B2C world.
