‍

‍A conversation with Arron Ritchie, Head of Data Science at Virtusize

‍

‍

Artificial intelligence (AI) is transforming industries across the board, but in fashion e-commerce the stakes are especially high. Unlike electronics or books, clothing has nuance: how it fits, how it feels, and how it reflects personal style. For e-commerce managers, getting these details right is the difference between loyal customers and costly returns.

To explore how AI is reshaping fashion retail, we sat down with Arron Ritchie, Head of Data Science at Virtusize. In this wide-ranging conversation, Arron discusses how AI is evolving, the biggest opportunities and pitfalls for fashion e-commerce, and why data readiness is the foundation for success.

The Rise of AI in E-Commerce

Q: How do you see the rise of AI impacting e-commerce broadly speaking?

Arron Ritchie: The rise of AI in e-commerce mirrors what’s happening across other industries. Early on, machine learning was mainly used for operational tasks like demand forecasting, warehouse optimization, and staffing efficiency. Those models gave businesses the ability to plan better, reduce costs, and spot inefficiencies they couldn’t see before.

The more recent boom in large language models (LLMs) like ChatGPT has changed the conversation. Suddenly, we’re not just talking about prediction models but about tools that can interact with humans, generate text, and even mimic reasoning. These tools have opened up exciting new possibilities – automating repetitive tasks, generating marketing copy that feels personalized, and integrating into workflows that used to take hours of human effort.

But the challenge is that out-of-the-box LLMs are good generalists. They’re like Swiss Army knives: they can do a lot of things, but they aren’t necessarily great at any one thing unless you add additional infrastructure. That infrastructure costs money, requires expertise, and takes time to get right. So the question for companies isn’t “Should I use AI?” It’s “Where does AI actually add value for me, and how do I tailor it to my needs?”

Opportunities for Fashion E-Commerce

Q: Specifically for fashion e-commerce, where do you see the biggest opportunities? Arron Ritchie: Fashion e-commerce has a very particular challenge: bridging the gap between in-store shopping and online shopping. In a store, you can try on a t-shirt, check how it feels, and make a confident decision. Online, you don’t have that experience — which is why returns are so high and why shoppers often hesitate before purchasing.

Digital fitting solutions like what we provide at Virtusize are one way of solving that, but they’re only the beginning. Every shopper is unique, with different body shapes, preferences, and styles. That’s where personalization comes in. With the right AI models, we can go beyond “You’re a Medium” and say, “This shirt will feel snug in the shoulders but loose at the chest.” That kind of insight builds confidence and makes online shopping feel closer to real-life shopping.

Generic AI tools can do some personalization, like recommending more black products to someone who’s bought black jeans. But fashion requires more sophistication. If you try to force a general-purpose AI into fashion without adapting it, you’ll either waste tokens, get inconsistent results, or end up with recommendations that don’t really resonate. Purpose-built models for fashion – grounded in real purchase data and user preferences – are where the real opportunities lie.

Beyond Fitting: Personalization and the Shopper Journey

Q: Beyond fitting, what else could fashion e-commerce companies do with AI?

Arron Ritchie: Personalization has huge potential that hasn’t been fully tapped yet. Right now, most brands only know what a customer does on their own site. But shoppers don’t live in walled gardens. They shop across brands, across categories, sometimes across industries. A customer might buy jackets from one brand, shirts from another, and jeans from a third. Yet when they land on your site, they get the same homepage as everyone else.

In a physical store, a good salesperson takes context into account. If they see you carrying three jackets, they’re not going to push more jackets on you. They’ll suggest trousers or a sweater to go with them. Online, that context is missing – and AI is the bridge to bringing it back.

Imagine tools that can correlate behavior across different websites, or even infer preferences from broader shopping patterns. A system that notices a shopper loves button-up shirts and prioritizes them on the homepage. Or one that detects their preference for certain colors or cuts and tailors the assortment accordingly. That’s the kind of intelligence that makes online shopping feel truly personal.

We’re not there yet as an industry. A few brands experiment with chatbots or simple recommendation engines, but very few are integrating broader data. Part of the reason is privacy – and part of it is just technical complexity. But the direction of travel is clear: more contextual, cross-brand personalization that goes beyond surface-level suggestions.

Internal Efficiencies: The Less Glamorous but More Impactful Side of AI

Q: Outside the shopper experience, what internal opportunities does AI offer e-commerce businesses?

Arron Ritchie: Internal efficiencies often have the biggest business impact, even if they’re not glamorous. Demand forecasting is a perfect example. If you can predict how many t-shirts to stock in Tokyo versus London, you avoid both stockouts and excess inventory. That’s real money saved.

Other areas include logistics, where route optimization can cut delivery costs, and staffing, where AI helps you schedule more effectively. CRM systems also benefit – AI can identify patterns in customer data that a human team might miss.

The problem is that these solutions don’t get the headlines. People get excited about chatbots and flashy LLMs because they feel new and interactive. But if you look at ROI, internal systems are just as critical. In fact, if you focus only on customer-facing AI and ignore the internal side, you’re missing half of the value.

The Data Challenge

Q: Many companies want to dive into AI but don’t feel “data ready.” What’s your perspective?

Arron Ritchie: Data is everything. Without clean, centralized, and reliable data, you’re building on sand. AI is like the furniture and décor of a house – the flashy part. But the foundation is your data. If the foundation is weak, the house collapses.

Too often, companies try to implement AI without fixing their data first. They have fragmented systems, duplicate records, or outdated databases. The result is poor performance and wasted money. Before you even think about AI, you should ask: Do we have a single source of truth for our data? Is it clean? Is it accessible?

If the answer is no, that’s your starting point. For some businesses, it makes sense to work with external vendors who already have clean, large-scale datasets. At Virtusize, we’ve spent years collecting purchase and fit data across hundreds of brands. That allows us to build accurate, scalable models that would take a new player years to replicate.

Virtusize’s Latest Developments

Q: What is Virtusize focusing on right now in terms of AI and machine learning?

Arron Ritchie: Our focus is making online shopping feel as close as possible to in-store shopping. We’ve always provided size recommendations, but we wanted to go deeper. Now, instead of just telling you “You’re a Medium,” we show you how the garment will fit: snug here, looser there. It’s about replicating that feeling of trying something on and thinking, “This is exactly how I like it.”

We recently rolled out an upgraded fit logic that uses machine learning to better balance body data with individual fit preferences. Our earlier models already did a solid job with body measurements, but with this new version, we’ve trained it on years of purchase data so it understands what “good fit” actually means – depending on the category, the brand, and the style. It’s a more adaptive, realistic approach to how people actually shop and wear clothes.

The results have been impressive. In A/B testing, we’ve seen 25-40% improvements in size recommendation accuracy. That means more shoppers are choosing the recommended size, buying with confidence, and keeping their items. It’s a big step forward – and it’s only possible because we had the data foundation in place first.

Practical Advice for E-Commerce Managers

Q: For e-commerce managers who want to start with AI, what’s your recommended action plan?

Arron Ritchie: First, don’t believe the hype that AI will solve everything. Start with a clear use case that addresses a real business pain point. If returns are your biggest issue, look at fitting solutions. If stockouts or overstocks are costing you, explore demand forecasting. If customer engagement is low, think about personalization.

Second, decide whether to build or buy. Building in-house only makes sense if you’re ready to invest for the long term, with a dedicated team and a multi-year plan. For most companies, the smarter path is to partner with vendors who already have the expertise and the data.

Third, set clear success metrics. Don’t just say, “We implemented AI.” Ask: Did it reduce returns by 10%? Did it increase conversions by 5%? Did it save hours of staff time each week? If you can’t measure it, you can’t justify it.

Finally, be cautious but curious. There are plenty of “snake oil” claims in the AI space – tools that promise to solve every problem. Good partners will be realistic, acknowledge limitations, and explain trade-offs. If it sounds too good to be true, it probably is.

Looking Ahead

Q: What excites you most about the future of AI in fashion e-commerce?

Arron Ritchie: What excites me is the unknowns. Every week, there are new breakthroughs – new architectures, new methods, new possibilities. Some of them will be overhyped, but eventually one will stick and become the next big leap forward.

Fashion retail is reaching a critical mass of data, which means we can start adapting ideas from other industries. Maybe it’s demand forecasting methods from logistics, or personalization strategies from media platforms. The point is, we now have enough data to experiment – and that’s when real innovation happens.

For me, the fun is in taking these new ideas and making them practical. It’s not about shiny demos or one-size-fits-all tools. It’s about solving real problems for shoppers and retailers, whether that’s reducing returns, increasing conversions, or making online shopping feel as easy as walking into a store.

Final Takeaways: Steps You Can Take Now

If you’re an e-commerce manager wondering where to start, here are practical steps you can take today:

1. Audit your business pain points.
   
   • Are returns your biggest issue? Explore fitting solutions.
   • Is inventory mismanagement costing you? Look at demand forecasting tools.
   • Is customer engagement low? Focus on personalization.
     
     
2. Assess your data readiness.
   
   • Do you have clean, centralized, usable data?
   • If not, prioritize fixing this before chasing advanced AI.
     
     
3. Start small and measurable.
   
   • Choose one use case to focus on.
   • Define clear success metrics (reduce returns by 10%, increase conversions by 5%, etc.).
     
     
4. Buy before you build.
   
   • Partner with proven vendors who bring both data and expertise.
   • Only consider in-house teams if you’re ready for a long-term investment.
     
     
5. Stay skeptical – but curious.
   
   • Avoid inflated claims that promise to solve everything.
   • Experiment, measure results, and scale what works.‍
     

Recap

AI is not a silver bullet, but it is a powerful set of tools. For fashion e-commerce, the winning strategies balance personalization with operational efficiency, built on a solid foundation of data. As Arron Ritchie emphasizes, the smartest approach is to start small, stay practical, and focus relentlessly on delivering measurable value for both customers and the business.