Striking the AI balance: How to maximise the opportunity while minimising the risk in global retail
Artificial intelligence is already playing a crucial role in the shifting online retail landscape. From personalised shopping advisors to semantic search, the technology is shaking up the ways we do business at the front end, back end, and everywhere in between.
At OnBuy we have invested fully in AI, which has been instrumental in the growth of our disruptive cashback marketplace model, allowing us to establish ourselves as a real challenger within the industry.
The time and cost savings AI has brought to the business are allowing OnBuy to grow further and faster than many thought possible. It has been a crucial enabler for us, giving us the capacity to double down on our cashback proposition. Thanks to its transformative effect on our profitability, we're now in a position to give back over £15m every year to our customers.
Having witnessed these gains ourselves, it's no surprise that so many retail businesses plan to invest in generative AI for their business in 2025.
Despite being fully invested in harnessing the benefits of AI for my business, I firmly believe that rushing to implement flashy features can backfire if not approached with caution.
Although AI offers e-commerce businesses never-before transformative potential, its challenges are also unprecedented. To get the most from AI, online retail must take a balanced approach to implementation – caution, consideration, and a customer focus are key.
Walk, don't run
There is a misconception that AI software is just plug-in and play, but there are a lot of complexities when it comes to implementing it successfully, especially if you want to provide a high-quality customer experience. Improving efficiency is important, but customer experience is paramount.
AI tools contain random, probabilistic output and therefore need to be monitored by humans, so the AI model doesn't drift from its intended purpose.
OnBuy contains a huge quantity of information. With millions of products sold across our site, we must have absolute confidence that anything we implement can perform at scale and not deteriorate in performance over longer-term use.
Being aware of the limitations of AI products is as important as accepting the possibilities. Any production AI patterns should have human involvement to make sure these systems produce consistent results and have sufficient guardrails where active decisions are being made. To use AI responsibly, businesses must do thorough research and shelter customers from inherent AI biases where possible.
Prioritise customer experience
You don't need me to tell you that the customer is the core of the online retail industry – but when it comes to implementing AI, too many businesses forget this simple fact in favour of innovation for innovation's sake.
Just because you can transform a process, product, or experience using AI, doesn't mean you should. Improved user experience should always be the end goal. Let customer data and needs guide your AI journey, ensuring the technology you choose works for those who matter most.
Engage stakeholders on your journey
Effectively implementing generative AI in the e-commerce space requires going beyond due diligence to inform and engage stakeholders from all sides.
Start small. Use AI to assist with general day-to-day tasks, like generating the correct formulas on Excel, or streamlining research data. As confidence grows in how to use AI and how beneficial it is to your business, then you can explore more advanced applications.
While AI features can positively shape your platform – whether improving inventory management or enhancing customer service provision – it's essential to keep human users at their heart. Both staff training and consumer education are required to ensure your investment pays off, retains the human touch, and delivers the results you were aiming for.
More than chatbots
One prominent example where a cautionary approach is vital is AI shopping assistants. We're seeing these pop up all over the online space – but just because everyone else is doing it, doesn't necessarily mean it's the right time for your business to hop on the trend.
For any e-commerce platform in the early stages of its AI journey, rushing to implement customer-facing features like chatbots can be a case of putting the cart before the horse, and poses a serious risk to ROI.
Chatbots are only as good as the data you use as context for text generation. They need a solid technological framework, business accountability, and the right data and continuous human training and monitoring. This ensures accuracy, functionality, security and customer appeal.
There's no harm in having shopping assistant bots on your AI roadmap, but they must be implemented correctly to ensure they add value to your business and enhance the customer experience.
The AI-assisted future of online shopping
At OnBuy, we're continuing our journey with AI with one philosophy in mind: we're an online cashback marketplace enhanced by AI, not an 'AI marketplace'.
We've already been using AI behind the scenes to reduce risk, protect from fraud, improve seller performance, and enhance customer service and product management.
With this strong foundation in place, we're making customer and seller experience our focus as we evolve and expand. As we adopt more AI across the business, it will help us offer better recommendations and an all-around more inspirational shopping experience.
AI is not a shortcut to success in online retail – it's a tool to enhance it. The key lies in a thoughtful, balanced approach: building strong foundations, aligning with customer needs, and staying true to your business goals. Before reaching straight for flashy, impressive AI tools that make headlines (sometimes for the wrong reasons), focus on solving back-end challenges to start perfecting seller and customer fundamentals.
By leveraging AI to solve real challenges and elevate experiences, we can unlock its transformative potential without losing sight of what matters most – the people we serve.
