AMA: Hot Takes on Headless, AI, and Personalization from an Ecommerce Consultant

Amy Sillince is a freelance luxury ecommerce consultant who has worked in the ecommerce industry for 15 years. She started her ecommerce career in the fashion sector in London. She has spent about 10 years in the Salesforce Commerce Cloud ecosystem — first as a client, then as a solution architect for the Commerce Cloud team. She has had her own consulting firm, Lux Commerce, since 2023, where she creates elegant online experiences for brands and partners in various industries, from fashion and retail to consumer healthcare.

We caught up with Amy to chat about hot topics in ecommerce: AI (we know you saw this comin’), personalization, and headless. Here’s our conversation, lightly edited for clarity and conciseness.

Q: How has AI changed the way you work, or the types of projects your clients want you to focus on?

A: Like everyone else, I’ve been having a lot of chats with friends and colleagues about AI recently. We can see how fast it’s changing the ways of working — and how our mindset needs to evolve with it. So far, AI has changed my job more in the sense that my clients are asking me to find out the best ways for them to use it. From what I’ve learnt so far I like the potential in data analysis and decision-making.

These days, we have so much data available to us: about our products, about our customers, about our operations. There’s too much for us to sift through efficiently, and it can end up just laying dormant. In its current state, I like the case for using AI to answer questions quickly and spot trends — in the way that a human couldn’t. Or maybe from an angle that a human wouldn’t see or get to as fast. 

There are a lot of technical or creative things that AI is going to be able to do for us, but we’re still going to need smart people to guide all those things, and I think we have a ways to go to achieve them at scale. I’m more interested in identifying companies who are building genuinely useful AI tools to help facilitate our work, than dreaming up ideas and solutions from scratch within our teams right now.

Q: In your experience, what are the “low hanging fruit” when it comes to improving ecommerce performance with AI? What’s the most effective, low-effort way to make it happen?

A: Honestly, I find a back to basics approach works best. We know that AI is really good at answering questions and finding patterns. So why not start there before getting into more complicated use cases? AI can find insights really quickly — especially ones that a human might miss. It all comes down to giving it the right data. Most businesses that are established with an online presence have enough data, but whether they have it organised enough is an entirely different question, right? 

You need to have your data connected enough to make it make sense. And that’s something AI could help with. You can be like, “Here’s my sales data, here’s my on-site search data, here’s some data from my email newsletters and from Google Analytics. Find me some patterns amongst all of this and make some suggestions about where I have opportunities in my business.” Tasking a human with something like that can be really heavy lifting and requires a lot of time, which can be used on other things.

Q: It’s difficult to talk about ecommerce data without also discussing personalization. Where should businesses start if they want to improve personalization?

A: Start by identifying what will be useful to your customer. That should always be your north star. Look at your data and observe users’ interactions with the site to find friction points in the customer journey. Or look for areas where you can improve sales metrics, like units per basket or average order value. Personalisation should be about enhancing the customer experience, not just making the website more sophisticated for the sake of it.

In fact, I think there are scenarios where you could actually even overcomplicate personalisation. I’ve seen this happen with merchandising on product listing pages or in the search experience, where AI can be over-sensitive to the user’s behaviour. If you’re a lean team, it’s important to investigate whether so many versions of personalised content are actually making an impact on sales, or whether the time and resources might be better spent on something else. 

I think there’s more courage in making the right decisions for your business and customers, than pushing something mostly because you see your peers or competitors doing it – do you know whether what they’re doing is really working?

Q: You’ve worked with a lot of companies to improve ecommerce personalization. What are some common pitfalls you’ve seen businesses make?

A: One common pitfall I see is pushing personalisation too far and making the experience too sensitive to user behavior. For example, a shopper might look at a list of shoes and click on the pair they like — but then they click on something else. As the site learns about the customer, based on where they click and browse, an oversensitive personalisation function might change the content or merchandising too quickly. When the shopper goes back to the first listing page where they initially saw the perfect pair of shoes, the page has already rearranged itself and the shoes have moved from where they found them. That’s a really annoying and disorienting experience. There needs to be a balance between personalisation and user experience to ensure that the shopper’s journey remains familiar and intuitive.

It’s important to have a softness to the experience. Imagine how frustrated you would be if you walked around a grocery store and everything had moved on the shelves since yesterday. Businesses should be creating personalisation that feels natural in the environment. 

On Commerce Cloud, that would mean blending rules together. For instance, you could blend some percentage or weighting of the personalisation function with some other rules that aren’t  AI-based. It’s important to have control over how heavily personalisation intelligence factors into the way things behave on your site. If you’re developing something from scratch, the balance between personalisation and UX should definitely be a discussion with the development team.

Q: You’ve mentioned that API-led architecture like headless or composable has been on a pedestal for a few years now. How do you think businesses should approach headless?

A: I come at headless from a different angle than a lot of people who talk about it because I approach it from more of a business standpoint, since that’s my domain and expertise, rather than from a developer or IT background. Both of those perspectives are necessary to have a productive headless conversation. But what I have seen so far is really that conversations about headless are typically very heavily dominated by IT teams — and the business tends to get forgotten. I have seen firsthand projects where teams are developing very cool, sophisticated headless solutions that tech teams are really excited about, but somewhere along the way they have probably forgotten to check whether it’s actually going to work for the business day to day.

So my advice is get all the right people in the room from the get-go. And if there are stakeholders who don’t really understand what headless is, it’s your job to explain it to them and make sure that they do. Successful headless projects require all invested stakeholders to be able to give their own truly educated feedback and opinion on their specific corners of the business.  

It’s also important to hear from some of the people who are hands-on in the platform you have today. The goal is to find out how they do their jobs and the way they’re using your current platform’s features. You don’t want to overlook or eliminate an important business process, or overcomplicate something that used to be simple for your teams. That’s not to say you shouldn’t change things. But unless you go and ask the hands-on users what they actually need to run your site and make sales every day, you could end up with a very cool, beautifully built headless website – that goes really fast and has super clean code – with a team that can’t run it or generate sales with it. Maybe that’s controversial, but I really believe that’s the truth.

Q: It’s a disruptive time in ecommerce (well, in every market and industry). What are the things that excite and invigorate you about the future of ecommerce? What, if anything, do you think businesses should be cautious about?

A: I’m really excited about how online shopping behaviours will continue to evolve – and how our jobs as ecommerce professionals will need to adapt and respond to that. It could be due to AI and agentic commerce, more connected shopping experiences thanks to headless, or just shifting behaviours and trends.

I’m also curious to see how the customer journey may change, and what different ways of thinking or new skills we will need to develop to support that. For me, the important thing will be to make sure we still show up where our customers are, and focus on what we can do to make their shopping experience as easy as possible. Focusing on the tech alone isn’t enough. We need to make sure we’re doing the best thing for our customers; without them, there would be no ecommerce.