Welcome to part two of an interview with Mark Evans, GM and VP of Fuel Pricing and Logistics at PDI Technologies. Read on to explore how data science and AI are shaping modern fuel pricing strategies around the globe. View part one here.
Q: How should businesses go about implementing AI and machine learning?
A: True differentiation comes from using AI strategically, in a way that aligns with your business model and adapts to real-world conditions. You have to understand the market and your primary competitors. Your biggest competition might not be the station across the street. It could be changing customer behaviors, alternative fuels, or even loyalty-driven pricing models.
You also need to carefully define your brand promise to your customers. Not every business competes on price alone. Some focus on service, brand loyalty, or customer convenience. Your pricing strategy should reflect what sets you apart, so any AI and machine learning tools should align with your long-term strategy and brand promise.
Q: How do you achieve that level of flexibility in your pricing strategy?
A: You must be able to adapt instantly in regard to daily market movements. Market conditions shift constantly, so AI gives you the capability to respond just as quickly.
To make sure that strategy works, you need to constantly measure your results. This is where a lot of businesses fall short, because they implement AI and assume it’s working. But AI isn’t magic. You need to test, refine, and validate it continuously.
Is AI-driven pricing improving your revenue or driving customer retention? Is it aligning with your business goals? You have to hold AI accountable just like any other pricing strategy.
Q: How do you ensure you truly understand the market versus just making decisions based on informed assumptions?
A: To achieve that level of clarity, I recommend focusing on three key areas: what truly drives pricing, whether you’re tracking the right competitors, and how you’re refining your approach over time.
AI can help you uncover insights and patterns that aren’t always obvious. For instance, how sensitive is demand at each site? A small price move at one location might cause a major shift, while another site sees no impact. AI can help you compare all nearby sites to reveal who actually influences your business, sometimes in ways you wouldn’t naturally expect.
AI makes it much easier to test different approaches, so you can quickly measure the outcomes, identify what’s working, and continually refine your strategy.
“You need to ask whether you’re feeding your AI the right inputs or just reinforcing old assumptions. If an AI is built on the wrong data, it will take you in the wrong direction.”
Q: Once you understand the marketing dynamics, what are the next steps to deploying AI or machine learning technologies?
A: It helps to build a machine learning model that not only helps you define prices but continuously learns, adapts, and improves over time. A strong model starts with good data, including historical sales, market trends, customer demographics, weather, traffic patterns, and local pricing.
Next, you need to clean, normalize, and structure your data to ensure accuracy, because bad data leads to bad decisions. Not every data point is relevant. Key variables—such as demand patterns, site characteristics, and competitor movements—help refine your model for better accuracy.
To develop your model, you need to select different algorithms—like regression models, decision trees, or neural networks—based on data complexity and business needs. Then you need to train your model and evaluate it to make sure it’s working as expected.
Lastly, you need to monitor your model through real-world testing and iterative refinement. You have to train your model with fresh data, and once you’ve deployed it, you can then use A/B testing to compare AI-driven decisions against control groups to ensure continuous optimization of your model.
“The ability to offer the right price, at the right time, to the right customer will be a key competitive advantage.”
Q: What’s next on the horizon for fuel pricing?
A: Over the last three decades, we’ve witnessed a significant technical evolution, migrating from simple rules-based systems to highly sophisticated, cloud-native, and AI-powered models. Today, most forward-thinking companies are already leveraging machine learning and dynamic pricing to react to complex market conditions in near real time
These AI-driven systems enable pricing to adjust dynamically based on granular demand patterns, competitor movements, and even specific customer behavioral data. This precision, the ability to deploy the right price at the right time for the right customer, is no longer a luxury. It’s the fundamental engine driving competitive advantage in modern fuel retailing
Q: What’s needed for that type of transformation?
A: A true transformation requires smarter data convergence. Pricing can no longer operate in isolation. The most significant business benefits will materialize from the full integration of pricing systems with critical platforms like ERP and logistics. This level of connectivity allows pricing to actively anticipate supply chain disruptions, optimize distribution costs, and drive overall operational efficiency. This tight, AI-driven convergence is what will truly create competitive separation.
The next technological leap will be truly agentic AI. We’re moving beyond the current reliance on static dashboards and reports. Instead, agentic AI will power proactive, autonomous systems capable of engaging with complex pricing scenarios in real time. For businesses, this will translate directly into a powerful competitive edge, meaning pricing decisions are not just optimized, but anticipatory.
Ultimately, this points to fuel pricing becoming more intelligent, more connected, and highly strategic. It’s all about enabling pricing, supply chains, and operations to function seamlessly as one synchronized engine to stay ahead of market dynamics.
Connect with Mark on LinkedIn.