Author: Tim Blake, Enterprise Sales Iberia and Italy at commercetools
From: “AI: The New Frontier in Retail” Shoptalk event
The meteoric rise of Generative AI (GenAI) is propelling us into the future of retail faster than ever before. But are you able to cut through its noise to embrace its transformative power, or are you at risk of being left behind on the shelves?
As we stand on the cusp of this exciting new tech revolution, Commercetools asked digital pioneers from Selfridges, British Gas, and Constructor to share their GenAI strategies to secure market differentiation, boost customer engagement, and unlock business success.
GenAI can feel like lightning in a bottle right now. How are brands unleashing its power across the market?
Fabrice Khullar, Director of CX, Product, Tech & UX at Selfridges, sums up its current use cases and benefits: “It's about using data to create something new.” That could mean creating images and videos, which would be more cost effective for a multi-brand retailer than working with suppliers. Or redesigning new websites or product pages more quickly.
For example, Charlotte Todd, Ecommerce & Marketing Director at British Gas, shared that she is exploring how GenAI could learn to churn out content to more specific audience demographics across relevant channels, working quickly and iteratively without going through weeks of an external production process.
“Where it really comes into play for us is ensuring that we are presenting the right content to the customer at the right time so that the journey itself is personalized based on their needs,” she explains.
Selfridges recently ran a trial which showcased the benefits of GenAI’s personalization capabilities. The retailer AB tested a carousel ad on the homepage of its app, using GenAI to personalize the ad creative to the user demographic. The result was a 130% uplift in the click-through rate (CTR).
The key, as Fabrice advises, is to understand where, when, and how to “apply the data, so it’s more meaningful, to drive more revenue.”
So, how should retailers kick off that process?
The hype around GenAI means there’s a danger that retailers get sucked into its noise, rather than thinking more strategically about what GenAI can do for them.
As Constructor’s founder & CEO Eli Finkelshteyn says: “There are a lot of solutions in search of problems in GenAI right now. And that’s just not the right way to build any kind of product. All of the focus that exists right now on building things that look cool misses that those same things may not actually be useful to customers.”
It’s important to figure out the problems that need solving before understanding if and how GenAI can solve them. That approach will result in a more meaningful understanding of customers, which will result in a stronger, longer lasting relationship with them.
The decision to build or buy is a question that’s existed in software for years. And now we’re starting to see the same dilemma when it comes to AI solutions.
Internal teams can use open source to create and train an AI model to a decent level of accuracy. But ultimately, our experts agree, external specialists will be able to do the job faster — with better results — as they have huge data sets that a single retailer won’t have access to.
“As a retailer, ask yourself: do you really need to hire data scientists when you could probably go and find a supplier who solves that problem really well?” Fabrice comments.
And when it comes to choosing a supplier, go back to your business objective, research the market, and check whether your shortlist is “really going to do something different and specialist enough that you need them to do it,” he adds, bearing in mind that retail ecommerce is a specialist sector that needs experts in purchasing, inventory, and so on.
Finally, our panel agrees you should verify the tech can be integrated into other existing technologies, as the combination could end up being a “game changer.”
Our experts are clear: when it comes to GenAI, just because you can do something, it doesn’t necessarily mean that you should. So, how should retailers decide which AI projects to undertake, and then, how should they measure success?
Eli shared his golden rule: “The goal of any early product should not be to be perfect. It should be to be genuinely useful to at least some of your users. You can perfect something that's useful later, but you'll have a much harder time getting users to engage with something they don't find useful, even if it does what it's supposed to do perfectly. You don’t have to make something that’s perfect for everybody – you just have to make something that’s useful for enough people.”
For example, it might be that the original objective is to offer customers a new product discovery solution like an AI chatbot that lets users make long natural language queries. A perfectly reasonable goal could be to target just a subset of your users and focus on just making the most tech-forward 2% of them happy enough to keep using it and to tell their friends about it so that next year, 5% of users use it, and the year after 15% use it.
The system does not have to be perfect to do that. It just has to be useful and convenient enough for those early users to build the momentum you're looking for. The key performance indicator is “am I getting to enough users fast enough? And am I showing them enough value so they're actually coming back,” advises Eli.
Want to learn more about how to navigate the complexities of emerging tech so you can transcend the AI hype and confidently make strategic, future-fit investments in ecommerce technology? Check out Constructor’s ebook, Forward. Fast. The Future of AI in Product Discovery.