AI sizing and fit experts in conversation for fashionUnited

AI agents are already transforming how consumers discover, evaluate and buy apparel. How can brands and retailers cut through the agentic hype and focus on what actually matters to their businesses now? The answer starts with an area often overlooked in the AI conversation: sizing and fit.

Alvanon is the world’s leading sizing technology company. In this edited conversation for FashionUnited, Don Howard, executive director for Alvanon’s global consultancy and Eric Lee, the company’s executive director for the Americas, deliver three practical takeaway actions to help businesses succeed in the agentic era.

Eric: We'd best start with a definition of AI…

Don: As a consumer - AI represents a transition from traditional search engines, like Google, where we enter parameters and get a list of results to filter through, to AI agents that search, filter and recommend the best match according to our prompts.

For general products that are dimensionally stable AI is incredibly effective. However, we work in the apparel industry with soft goods that present a unique set of challenges. Chief among these is sizing and fit. We can engineer garments to a point, but all garments behave differently.

Eric: By way of example, I asked AI to "Give me three options of shirts between $70 and $100." In seconds it gave me three options, brands I'd never bought from before. They all looked perfect. So I asked it to do a virtual try-on for me, and it simulated all three shirts on my body, using publicly available information. I chose one and asked it to recommend the best size. After a couple of questions, it suggested a medium/slim, which I bought. When it arrived the reality was very different to the virtual try-on!

Eric: This really illustrates the disconnect between what an AI agent recommends and the consumer's experience. AI can be brilliant at directing first-time shoppers to a brand, but it can also create a big spike in returns and undermine consumer confidence. This is a costly disconnect. The truth is AI shopping tools can really boost new customer acquisition, but potentially at the cost of customer lifetime value - what a waste.

If I was a $100million ecomms shop and can reduce my returns by 1%,  that's  $1m back in my pocket! But a 5% spike in returns would have a huge impact on my bottom line....  

Don: Apparel fit is still the number one reason consumers return garments. The question to ask is, ‘what's making a garment return at a higher rate?’ Citing ‘fit’ as the reason for a return is too vague to be helpful in fixing the problem. We have to go deeper than the word 'fit' to understand why a garment has been returned. And brands need to ask ‘is our sizing and fit correct in the first place?’ 

Eric: That brings us to the first of our three practical takeaways: How do we help AI agents find our products as opposed to our competitors?

Don: Get your sizing right. Get it locked in consistently and then communicate it clearly across your teams, suppliers and ecommerce. It sounds simple but too many brands get this wrong.  Size and fit are not the same thing. Size is a brand’s defined, engineered core standard and the foundation of all garment designs. Fit is a condition applied over that size. So if a brand’s core size isn't standardised and locked in, there will be trouble ahead.  Brands also need to understand how body size and shape change as the garments change size. Too often, what is executed in production and experienced by the consumer, doesn't match what is promised online.

Eric: Absolutely. In my previous ecommerce role, I worked on photo shoots. We'd have a rack of clothes and sometimes wouldn’t know if the fit intent was supposed to be slim fit or relaxed. It illustrates how different teams have different priorities. A tech design team is looking at this through a different lens versus an ecomms team on a photo shoot focused on optimising conversion rates. This aligns to your point about communication of sizing and fit intent. A size chart is nice but it is just one piece of a larger puzzle. People return stuff because ‘it doesn’t fit as expected’, which is different to ‘I bought the wrong size’. It’s a subtle but important difference - consistency in sizing is the key to solving both.

Don: The size chart must be accurate and aligned to what a brand actually produces.  AI will read the size chart and other publicly available information such as consumer reviews with useful fit feedback.  If a brand changes its fit model for example for size or shape, the garment may look accurate online while producing very different end results and potentially more returns. 

Eric: Takeaway number two: How can brands analyse the data generated by ecommerce?

Don: Data is super important - return rates, reasons for returns, sales by size and more. The data shines a light on the consumer experience and response to what the brand has executed. 

Analysis should be actioned carefully and quickly! If there’s a spike in returns due to garments being ‘too big’, the body standard could be a factor but should not be the first question - rather ask, ‘was the garment made intentionally big and not matched customer expectations?’ Limited focus on one metric can lead to very different decisions without doing an in-depth analysis.

Eric: AI agents will read garment information very quickly so brands must have accurate size charts and fit descriptors in place on the product detail page. And the garments must consistently conform to these standards and descriptors or shoppers will be disappointed. Should brands have dedicated resources to analysing this kind of data? 

Don: Yes. Companies that do this well, typically have someone who understands both the commercial side and the technical sizing and fit intent of each garment. 

Eric: To our third and final takeaway: How do brands close the communications gap between different teams and silos, like ecomms and product development?

Don: This is probably the biggest opportunity. Historically these teams have existed in separate workflows with different responsibilities and timelines. Often the only interaction they have is when something is negative - a spike in returns or poor sales.  A more proactive approach to analysing data can help teams identify opportunities, from improving fit to developing product extensions such as tall, petite, plus size etc. Breaking down silos and being more collaborative and proactive is something most brands can do better.

The good news is that in the last 18 months or so, we’ve had many more collaborative conversations with ecommerce teams and cross functional leaders. The industry is recognising that we need to break down legacy