Everyone's talking about using AI to write product descriptions and create social media graphics. But according to three eCommerce experts who gathered for a recent webinar, that's not where the money is.
Host Chloe Thomas kicked off the eCommerce Explored session with a simple question: "Where are you seeing brands make the biggest AI implementation wins?" The answers surprised even the seasoned professionals on the call.
Meet the Panel
- Yaël de Keijzer is a content marketer at DooFinder, which builds AI-powered search and discovery tools for online shops.
- Judith van Dijk has over 15 years in eCommerce, working with BigCommerce, Shopline, and Shopify, plus helping small businesses scale through Enterprise Nation.
- Janis Thomas is an eCommerce and subscriptions specialist whose career spans from Birchbox to organizing Hugh Hefner's 80th birthday party.
Forget Content Creation - The Real Wins Are Elsewhere
Judith jumps straight in with an answer that goes against the grain: "I think the biggest wins are being made around content, and really that's a bit of a no-brainer, I imagine. I don't really want to focus on that side."
Instead, she points to customer service automation as the overlooked goldmine. "This is an area I think that still can be improved tremendously and is a lot overlooked because we spend a lot of time on the pre-purchase phase rather than the post-purchase phase."
She's talking about the unglamorous stuff: refund requests, lost parcels, basic product questions. These queries eat up hours of team time, yet most brands are still handling them manually or with clunky FAQ bots.
The difference with modern AI tools? "You can actually have a conversation rather than sometimes it's frustrating," Judith explains. "You answer a question like, it hasn't quite... That's just a standard pull out of my FAQs. I really have slightly different question to that."
These tools can consolidate emails, chat messages, and texts into a single dashboard, then provide intelligent responses that genuinely feel like conversations. Some platforms now have this built right in.
Yaël adds another layer that receives even less attention: AI working behind the scenes, which customers never see.
"I used to work at an eCommerce before, and I spent my days typing in a lot of different tags, a lot of different synonyms for the same product, and this could take a lot of time," she recalls. "But now using AI, for example, we have an AI synonym product that recognizes synonyms for the same tag. So you don't have to type all of these things in manually anymore, which takes hours of work."
Even better, AI can now analyze product images and automatically extract details, such as what it is, its color, and its materials. "You just upload an image of your product, and from that image, the AI can already extract what the product is, what the color is, what the materials are, et cetera," Yaël explains.
Chloe makes a sharp observation: "I think one of the reasons why these better implementations than creating copy, creating graphics is taking a while to filter through, it's because you've got to have the data in place and accessible to make it possible for the AI to be able to access it."
She's spot on. The power isn't in the shiny front-end tools - it's in fixing your database once and having everything flow from there.
But Wait - Content Creation Still Matters (Just Not How You Think)
Janis offers a different perspective that's equally important. She's not dismissing content creation - she's reframing what it enables.
"I think where it's really interesting is what it's evolving, and what it's freeing up time for, which is to realize things that you previously only dreamed of," she says. "You can take a flat product image and turn it into a lifestyle shot, that previously that would have been like, oh, I don't have time to shoot it, or get the props in, or do all of those things, whereas actually you can do it. You can turn a series of flat images into a video. You can get text to voice."
She points to a trend delivering real revenue: hyper-personalized creative at scale. "Particularly with Meta, we've gone from the half a dozen creatives to hundreds of creatives, and really where a lot of brands are seeing incremental revenue is shifting that to almost one-to-one creative personalization."
But here's the catch, and Chloe nails it: "Now the sky's the limit on what we could do with the creative, but we've got to put the inputs into it right, otherwise we get bad creative."
When you're feeding that content to Meta or Google, bad imagery trains their algorithms in the entirely wrong direction. "Strategy is the bit that might be missing," Chloe adds.
This leads to something both Janis and Judith emphasize: AI is forcing brands to get properly organized.
"Everyone's got a brand Bible that probably sits in a drawer somewhere and nobody looks at, and it's probably not fit for purpose," Janis says. "Actually, what different teams, even within the same organization, are doing is not aligned."
Judith agrees: "We're in a phase at the moment which is actually encouraging a lot of organizations to get their house in order as well... What AI is making us do is really think about, okay, what is the brand Bible that you have to upload? What are the FAQs? What are the product descriptions? And that all has to be done very systematically and properly."
She mentions Canva as an example: "You can upload your whole brand Bible. You can put everything in there so that it's constantly feeding off that information for any creative that you want to make. And that's such a fantastic exercise."
The Peak Season Question: Is It Too Late to Make Changes?
Chloe poses a question that's probably on everyone's mind: it's early October, Black Friday is weeks away - is there still time to optimize systems?
The panel splits on this one, and their disagreement reveals something important about how different businesses should approach AI.
Yaël is firmly in the "do it now" camp: "AI implementation doesn't mean you need to have a huge tech team on board on your company, and you need to spend hours on customization before it works. A lot of these content management systems, they have literal plug and play modules that you can implement and they use AI, and they can improve processes and systems in your company within minutes."
She's frustrated by businesses putting things off: "It's a shame that I see so many stores postpone things because they think it's going to be a huge project that they'd have to take up. But it's actually something that can be done super quickly in many cases."
Janis agrees and takes it further: "There's certainly, if you didn't feel that your performance is where you would hoped it had been, then you might say, actually, do you know what? I've been a little bit... experimenting a little with our button, how buttons look, and it's like, do you know what? If we think that we've made a mistake with those buttons, it will be much easier to change now than it would have been in previous years."
When Chloe pushes her - would you really make changes during Black Friday week itself? - Janis doesn't back down: "I think there's certainly... if you didn't feel that your performance is where you would hoped it had been, then you might" make changes even during the rush.
But Judith throws cold water on this enthusiasm: "I actually sit on the side of caution."
She explains: "I'd be very nervous to start testing new tools coming up to, coming right up to Black Friday, because a lot can go wrong. And when that stress hits and you're trying to deal with just that additional stress, you don't want to have the additional stress of how does this tool work? Why isn't it doing what it's supposed to be doing? I think that's quite a dangerous area to be in."
Her timeline is more precise: "We're about two months away, I think, just under two months away... But yeah, coming up to three to four weeks before Black Friday, I would be very cautious in launching new tools."
The difference often comes down to the size of your team and its technical expertise. "Are you a merchant, a brand that you're a tiny team, you don't have a lot of in-house tech experience, potentially? Or are you a larger company that has dedicated engineers sitting on the team? There's a difference there as well," Judith notes.
Chloe sums up the consensus: "I think we're in the 'it depends' arena, aren't we? It depends on the project, it depends on the team, and it depends on current performance."
But everyone agrees on one thing: "We are no longer in the 'we locked down the tech stack for four months at the end of the year,'" Chloe says. That era is over.
The Question Everyone's Thinking About: ChatGPT Shopping
Then comes the elephant in the room. Recent announcements about shopping through ChatGPT (with Shopify and Stripe) and Google (with PayPal and Apt) have everyone wondering: Is the website dead?
"I don't believe it is. I think the website is still going to be existing. I'll put that out there right now," Chloe says upfront. "But is this now going to become more of a marketplace? Is this going to become a big game changer?"
Judith tackles it from a data perspective: "I think it all eventually comes down to where is everything being hosted? I think that at the end of the day, there's still data that has to sit somewhere. And I don't see that in this conversation yet. Who are the big data players?"
Her point is crucial: "If you've got access to the data, what sits in front of that can be anything. It could be your web shop, it could be chat, it could be Google."
She even suggests eCommerce platforms like Shopify and BigCommerce might face "a lot of competition from tools or concepts that come out of nowhere. Because they look at where is the data being hosted."
Chloe draws a parallel to Amazon: "Amazon turned retailers, for want of a word, into product specialists. You know you're good at your product and you're good at dispatching your product... But then Amazon takes care of all the rest of it."
We might be heading toward another group of eCommerce businesses who are "data specialists" - brilliant at product information and fulfillment, but not particularly interested in their own website traffic.
Janis offers a fascinating comparison to streaming services: "We all thought streaming would be like, oh, there's one platform and you can watch everything. But instead everyone's got different subscriptions to Netflix and Apple TV and Amazon and all of those things."
Shopping through AI might work similarly - exclusive deals across different platforms create fragmentation rather than a unified system.
She also points out that consumer behavior changes slowly: "One of the things that I find, I've had some really interesting conversations in the last few weeks around the number of people who work in eCommerce that still use their desktop for certain purchases... There are people who even in their 20s and 30s say, 'Oh, if I'm buying a flight, though, I like to do it on my desktop.'"
If we still experience that disconnect with mobile shopping after decades, AI shopping adoption will likely be gradual as well.
The panel's conclusion? Commodity products, such as batteries, might quickly shift to AI assistants. "I'm pretty confident that if ChatGPT recommends me a AA battery, it's going to be an AA battery," Chloe says.
But complex, emotional purchases? "If I'm trying to find the perfect dress for Christmas lunch... I'm less likely to rely on ChatGPT. I may use that, then go through to the website to check all the details," she explains.
We're likely to progress through three stages: not using AI at all, using AI, then visiting the website, and buying altogether through AI. Different products and customers will progress through various stages over time.
The Bigger Shift: It's a Mindset Thing
Chloe throws out a statement for debate: "It doesn't matter whether you automate or turn to AI. It's about building efficiency and profits into your business. It's a mindset shift."
Her point hits home. eCommerce has lagged other industries in pursuing efficiency. "I'm finding a lot of brands talk about their AI benefits, but often it's actually automation rather than actual AI," she says. "I think we've just been far more focused on growth, growth, growth, and ignored the head office team."
Judith agrees: "I think that we're still in a little bit of a phase where when we talk about AI, we're talking about automation at the moment. And I do think that there is absolutely this opportunity now to probably stop and really have a look at how can you be more efficient and more effective with the way that you work."
She's seen this pattern repeatedly with smaller companies wanting to scale: "Often you'll see that it's the, and it sounds bad, right? It's the lack of resources and the lack of automation that's stopping them from actually scaling."
The instinct is to hire more people. However, Judith argues that automation has long been a better answer, and AI is taking it further by connecting everything. "It's not just achieving some efficiencies here and here and here, but it's connecting the whole lot, and that's creating a huge, big efficiency."
Though she admits: "It's complicated... Connecting all those separate bits is still very much in the realm of engineering and techies, and we're made to believe that you and I can do it." No-code isn't quite as simple as advertised.
Yaël makes an important point: "When we're talking about AI, they're also very connected to automation. It's a conversation that I have a hard time separating, if that makes sense."
She's right - the distinction often doesn't matter. What matters is identifying your problem and finding the best solution, whether that solution is AI-powered or not.
Janis highlights a crucial cultural shift in how teams need to work: "There's huge amount of stuff that just isn't answered by their FAQs or their site. Actually, it's a big mindset shift in terms of what their customer services team need to do. They need to spend less time actually answering queries and more time analyzing those queries and working out how to make sure that data then goes into that repository."
She's blunt about what happens if you don't: "If we just stay doing what we're doing, then it's going to be like, 'Oh, well, the AI CS bot is rubbish.' It's like, yes, because it doesn't have the material it needs to get better."
Chloe shares a brilliant example of a B2B company that built a chatbot but deployed it internally first. Their customer service team got AI-generated answers they could edit before sending. "Response times plummeted, team members handled more chats simultaneously, and the chatbot improved through observing human edits."
However, it took six months of data work before the system went live to customers. That's the reality - AI isn't instant magic.
Three Actions You Can Actually Take
As the session wrapped up, Chloe asked each panelist for one practical must-do action.
Judith's advice: Stop rushing into tools and plan properly first.
"Where we stumbled out of the start blocks was like diving headlong into problems. That's like, we have to do this. So you pick up the first thing that comes across your desk or pops into your head. And I don't think that's the right way to approach an AI transformation," she says.
Instead: "Step back and have a good long think about what does that process look like? Whether it's on the front-end for your customers, what's the journey they're going through? Where can we improve, or whether it's on the back-end side in your teams, how do we work? How do we bring a campaign to fruition? These are all the steps. Okay, where do I need to build the basis, the data, etc. And what parts of AI can I use to optimize that? I think that planning bit is so important."
Janis's advice: Get your teams aligned, especially on how they're using AI for content.
"I think you need to be bringing your teams together to at least make sure your prompts are being done in a consistent way," she says.
Her strategic point: "What should you be focusing on right now? Because all the things that we've discussed, there are a million things you could be doing. And I think particularly in this run up to peak, I think the strategic thing of what are we focusing on is really important."
Yaël's advice: Experience your own site as a customer would.
"Look at your own website and look at the search bar yourself. Go through your website as a user, as a visitor. Sometimes we think we have the products, we have the website is live, people are coming on our website, we're ready. But really also look if people can find those products in a way that is natural for people to look for them."
And crucially: "Do you have your analytics in place? Do you know what products people are looking for, how they're looking for them? Because this data is very important for improving your processes as well."
She sums it up perfectly: "Don't assume you're ever ready. Really always keep doing research. Always keep thinking from the user's perspective as well, where you can make these wins."
The Bottom Line
AI and automation aren't about replacing human judgment; they're about freeing people from repetitive work so they can focus on what truly matters.
The brands winning in 2025 won't be the ones with the most AI tools. They'll be the ones who've identified real problems, organized their data correctly, and implemented targeted solutions that actually solve those problems.
There's still time before peak to make meaningful changes, but it requires a strategy over hype. Get your data organized, align your teams on how they'll use these tools, choose solutions for actual problems you face, and test everything properly.
The benefits will last far beyond the Christmas season.
Watch the full webinar replay on YouTube to hear more from Yaël, Judith, and Janis. Want expert advice like this delivered regularly? Sign up for future sessions with the eCommerce Explored webinar series.
