Alessandro Mori

The brief looked familiar.

A mid-sized fashion brand was struggling with customers not finding what they were looking for: their bounce rate for visitors coming from certain creatives was skyrocketing; others were dropping off on the PLP, or after running a search; the take rate on PDP product recommendations was abysmal.

The brief was clear: review the entire upper funnel, identify opportunities to improve the UX, and execute.

So we did. We mapped their user flows, audited the product discovery UX, looked at their heatmaps. We identified friction points and proposed solutions.

But as we got to work, we started an uphill battle against an awfully maintained product catalog:

- Shoes were organized by gender, while apparel and accessories were unisex. A customer clicking on "Women" landed on footwear and found no path to jackets. A customer clicking on "Jackets" couldn't filter by gender.
- Attributes were filled in unevenly: different names for the same color, some products with detailed specs and others with none, variant names that followed internal naming conventions rather than how the customer thinks.
- A lot of the attributes customers want to filter or search by—occasion, fit, fabric—were not being captured in a structured format at all. There was no data to filter on, so there was no filter to build.

No matter how beautiful we made their search, filters, or recommendations, the data powering these features was too scarce and inconsistent to produce good results. Every improvement we made felt like putting lipstick on a pig.

And while we could have just shipped the cosmetic improvements and called it a day—that's what the brief asked for, after all—we thought they deserved better. So we went back to the drawing board: we scheduled an all-hands-on-deck workshop, involving everyone from their team who was responsible for or impacted by the structure of the catalog, and figured out a richer, cleaner structure that would allow the new UX to really shine. It wasn't easy, and it went from a one-month project to a multi-month nightmare of coordination, attribute enrichment, and data migrations. But it was worth it.

In systems thinking, you'd call the catalog structure a _leverage point_—the bottleneck that prevents everything else from working correctly. If you only work downstream of the leverage point, you will have minimal impact. But if you re-focus your efforts on the leverage point instead, that can unlock results you couldn't have dreamt of before.

Back to the project: our initial UX work wasn't wasted, but it was work at the wrong leverage point. The real problem was in the taxonomy, the attribute schema, the naming conventions. Once we helped the client restructure those around how customers actually browse, the discovery experience became much easier to solve. The interface hadn't changed dramatically—it just finally had something coherent to represent.

This is, we've come to believe, what CRO work looks like when it's doing its real job. Not shipping UI improvements, but using the discipline of studying how customers interact with a store as a diagnostic lens—one that, if you follow it honestly, keeps leading you to the real source of the problem.

## Finding Leverage Is Only Half the Work

Unfortunately, not every project has a happy ending. Sometimes, you find the leverage point and still can't get anyone to act on it.

Not that long ago, a subscription brand came to us with a retention problem. Their churn was insane, with most subscribers canceling by month six, and they wanted some help figuring out a strategy.

In our very first conversation with them, we threw around some ideas: redesigning the cancellation flow with better copy and deflection offers; launching a loyalty program to incentivize customers to stay; building new CRM flows to remind customers of all the value they were receiving.

But when we dug more into the data, we realized they had a deeper problem: as is often the case with subscription brands, most customers were just canceling because they had too much product.

You see, this particular brand had built a subscription model around regular shipments of a single, durable product category. By month three, most subscribers felt like they had more than they would ever use. By month six, what started out as a fun monthly treat had become an expensive annoyance.

Just like in our fashion project, working on the cancellation flow, loyalty offer, or product education wouldn't do anything to fix the fundamental misalignment between what the brand offered and what the customer needed. And just like in our fashion project, the work wasn't useless, but it was not aimed at the leverage point.

As we continued our data exploration, we realized there was a strong opportunity to rethink the subscription model itself, moving from a single-product cadence to a membership that gave customers access to a much broader catalog, giving them more ways to spend with the brand on their own terms. Less accumulation, more agency.

But this was a _very_ different project from the one we had been hired to do. What had started as a conversation with the Director of E-commerce now involved the CEO, the CMO, merchandising, logistics, and a willingness to rethink the business model from the ground up. The project was also very risky: while they already had the product they needed to support our proposal, everything else was a bet placed by a business that was already struggling. Yes, it could have turned the business around, but it could also have dug them an even deeper hole.

We built the case and presented it. They listened, asked many good questions, and decided they were fine with the problems they had.

This is the struggle with leverage points: at the end of the day, finding them is not enough. Because they're so cross-cutting, you will still need to make your case to people whose work will be affected. Some of these people may have never even asked for your opinion in the first place, and some of them may prefer things to stay the way they are, for a million reasons that you'll find very hard to understand. The deeper the intervention, the wider the coalition you need, and the more political the conversation becomes.

## Becoming a Better Player

While there's no formula for getting people to act on leverage points, there are a few things we've learned—some of them the hard way.

The place to start is almost always the org chart, not the system. In the subscription project, we spent our opening conversations with the Director of Ecom. That made sense given how the brief was scoped. But the decision we were ultimately asking for required many, many more people—people who hadn't been part of the conversation and weren't that eager to be involved. By the time we figured that out, we were already in the wrong room. Before you start making the case, figure out who the actual decision-maker is, who influences them, and what their agenda is.

Once you know that, don't limit yourself to data, but build a story instead. When you're asking for something this uncomfortable, people need to be able to picture the other side and see a real account of what life looks like once the work is done. How would a more flexible membership model fit within the lives of our customers? What cool things can we do with product recommendations once the data is actually clean? The clearer that picture, the easier it is for someone to imagine saying yes.

But while the story serves as your destination, you still need to figure out the smallest thing you can propose that actually proves your point. A full business model overhaul is a hard sell for a business that's already struggling (ask us how we know). A more contained bet, like offering a broader catalog to a small cohort of at-risk subscribers, would have been a much harder ask to refuse. And if it had worked, the bigger case would have made itself. The more cross-cutting the change, the more valuable a small proof point becomes.

To be clear, none of these will guarantee victory. Organizations have their own, opaque logic, and sometimes—alright, most of the time—that logic wins. This can be frustrating: we have been in more than one room where we felt like we were screaming into the void, while we could have just laid back and done what the brief was asking for.

Despite all this, it's still a battle worth fighting. The whole e-commerce industry is, by its very nature, extremely focused on the short term. We want results, and we want them yesterday. That's what makes UX and CRO improvements so appealing—and granted, they do have their place.

But at the same time, investing in cosmetic improvements in the presence of much deeper problems can lead to a lot of wheelspinning. When that's the case, you need someone with the ability to take a deep breath, find the one thing that matter, and focus all their efforts on it.

If this feels like the kind of work your brand needs to do, [perhaps we should have a chat.](https://nebulab.com/contact-us)

This Is Why Your E-Commerce CRO Program Isn't Working

Michela Frecchiami

Would you buy a $2,000 cashmere sweater without understanding how the fabric drapes on your body? Would you commit to a $5,000 handbag if you couldn't see the grain of the leather or how it sits on your shoulder? For most customers, the answer is no—and that's the fundamental challenge luxury brands face online. A poorly rendered material or unrealistic fit doesn't just lose a sale—it damages brand perception in ways that take years to rebuild. 

AI-driven virtual try-on is changing this equation. Augmented reality technology has existed for over a decade, but it's only now—thanks to its combination with advanced AI—that it's reached a quality level suitable for luxury. What's changed is that recent advances in AI, particularly generative models and diffusion algorithms, have transformed AR from a novelty into a genuinely useful tool. Today's systems can render materials with near-perfect accuracy, understand body types and movements, and account for real-world lighting conditions. 

But here's the critical insight: only a seamless, intuitive, and realistic experience can actually deliver benefits. The key is **realism over perfection**—virtual try-on needs to look authentic, not artificially polished. When rendering feels too perfect or synthetic, it triggers skepticism. AR models that slow down navigation, feel clunky to use, or produce artificial-looking results create a boomerang effect—damaging brand perception instead of enhancing it.

## The Business Case: Numbers That Actually Matter

Luxury and jewelry e-commerce typically sees conversion rates of around 1.0%–1.2%, which is significantly lower than the global average of 1.4%, according to Shopify's 2024 benchmarks. Yet retailers implementing AI-powered virtual try-on and 3D/AR experiences see consistent improvements: Shopify reports show that interacting with a 3D model increases purchase likelihood by 27%, while AR interactions can drive up to a 65% lift in conversion rates. 

More importantly, returns decrease measurably—Shopify case studies report 5–10% lower return rates on products using AR visualization, while third-party aggregated studies note reductions of up to 30–40% in specific categories. For a brand selling €800 shoes or €3,000 handbags, this doesn't just reduce logistics costs—it preserves inventory value and the ability to resell items at full price. 

Consider a brand doing €50M in annual e-commerce revenue. A 40% conversion lift on a 1.0% baseline brings you to roughly 1.4%—that's potentially millions in additional revenue. A 20–35% reduction in returns on high-ticket items can save hundreds of thousands of dollars in reverse logistics and preserve margins by ensuring products are sold at full price instead of being returned, reprocessed, and potentially discounted.

## How Leading Brands Are Implementing Virtual Try-On

The brands getting this right understand that virtual try-on must embody the same principles of realism over perfection that define their physical products. Here's how leading luxury houses are approaching implementation.

![](https://cdn.sanity.io/images/kqgmf0g2/production/ac99fdc24e43fdd74d8305bb7676cff3de11dc95-1999x1070.png?fit=clip)

**Gucci** has enabled sophisticated try-on across sneakers, accessories, watches, and beauty products with meticulous attention to material authenticity. Their AR rendering accounts for ambient lighting in real-time without looking artificially perfect—the leather on a Gucci loafer reflects light differently across its surface, just as it would in person. For beauty, Gucci offers virtual try-on of lipsticks and makeup through its dedicated beauty VTO platform on mobile devices.

**Chanel** focuses on beauty and accessories, enabling customers to virtually apply lipsticks and eyeshadows with remarkable color accuracy across different skin tones and lighting conditions. This is realism over perfection in practice: Chanel's renderings preserve real skin texture rather than over-smoothing it into synthetic perfection. For eyewear, their try-on feature captures a precise fit for different face shapes—critical in a category where how frames sit on your nose bridge or temples can make or break a purchase decision. 

**Luxottica** has explored multiple angles across its eyewear portfolio, including mobile apps, in-store kiosks, and smart glasses partnerships. Eyewear is particularly well-suited for virtual try-on, as customers need to see how frames complement their face shape, skin tone, and personal style. Their in-store kiosks let customers rapidly try on dozens of frame styles, demonstrating that digital try-on doesn't replace physical retail—it enhances it.

## What It Takes to Implement Successfully

The brands that succeed with virtual try-on share three non-negotiable principles: material fidelity that captures how products actually behave, interface design that feels native to the brand, and seamless integration across the entire customer journey. These aren't optional refinements—they're the foundation that separates implementations that strengthen brand equity from those that damage it.

Rendering quality separates successful implementations from those that damage brand perception. This requires extensive 3D modeling using photogrammetry techniques that capture products from hundreds of angles. Creating these digital assets properly requires significant upfront investment: specialized equipment, skilled 3D artists, and weeks of work per product line. The goal isn't perfection—it's fidelity. 

![](https://cdn.sanity.io/images/kqgmf0g2/production/9539e7101760bf9573222e41c2f94e2d9a5301dd-1600x1052.jpg?fit=clip)

Over-polished lighting, uniform textures, and idealized proportions tend to look synthetic. Imagine a cashmere sweater rendered with perfectly even lighting and no variation in the knit—it looks like CGI, not luxury. Strong implementations capture how products behave in the real world: the way leather reflects light differently across its surface, the natural movement of knitwear on the body, and the visible nuance of shades across different skin tones. That credibility is what distinguishes luxury-grade virtual try-on from mass-market implementations.

These digital assets become valuable inventory that can be reused across try-on, product pages, and marketing campaigns. However, there's no shortcut—low-quality 3D models produce experiences that are worse than having no try-on at all.

### Interface Design Must Align with Brand Language

High-quality 3D assets are necessary but insufficient. The interface itself must translate brand codes into interaction patterns that feel native to the luxury experience—not bolted on as an afterthought.

The interface should translate brand codes into the practical mechanics of try-on:

- **For minimalist brands**: Restrained UI with discreet prompts, few controls, and visual focus on the product
- **For more expressive houses**: Editorial layers and styling context, provided they don't distract from the core decision
- **Path to action**: One-tap camera activation, effortless switching between sizes, colors, and finishes, and simple save-to-wishlist or share-to-consultation flows
- **Continuity with the rest of the site**: Typography, tone of voice, motion, and microcopy should all feel native
- **Interaction patterns**: Mirror what customers already know from your PDPs and checkout to reduce cognitive load
- **Trust signals**: Explicit permissions for camera access, visible loading feedback, and gentle haptics that confirm key actions without breaking immersion

Above all, be selective: features like social sharing or gamification should only appear when they advance the brand's intent. In luxury, sophistication is often about what you remove; keeping the surface area clean allows the craftsmanship, materials, and fit to take center stage, making the experience feel effortlessly premium.

### Mobile Performance Is Mission-Critical

Mobile drives the majority of luxury e-commerce traffic. Performance is non-negotiable. Virtual try-on on mobile presents unique challenges: smaller screens, varied device capabilities, inconsistent camera quality, and suboptimal lighting conditions. The experience must be even more refined on mobile than on desktop, with rendering quality that cannot degrade—a luxury customer browsing on an iPhone expects the same material fidelity they'd see on a desktop.

Mobile-first design means rethinking the entire flow:

- Load times under two seconds
- Seamless camera access
- Flawless one-handed navigation
- Fewer steps to activate the camera
- Larger touch targets
- Simplified controls
- Aggressive optimization for bandwidth constraints

Most implementations fail here—the technology works beautifully in controlled desktop environments, but feels clunky and slow in real-world conditions on actual mobile devices.

### The Customer Journey Must Be Unified

Virtual try-on doesn't exist in isolation—it needs to connect seamlessly across every touchpoint where customers make decisions or seek reassurance.

Virtual try-on should sit exactly where decisions are made and then carry through the journey without friction:

- **On product pages**: Entry point should be prominent but unobtrusive, with states and preferences saved to wishlists so customers can resume their shopping later or discuss their purchases with a sales associate
- **During virtual consultations**: Associates should be able to reference the same try-on view to offer precise fit and styling advice, recreating the intimacy of a boutique experience at a distance
- **In-store**: Pre-built shortlists from at-home sessions can accelerate fittings and boost confidence
- **Behind the scenes**: Tying try-on engagement to personalization, size guidance, and CRM—always with explicit consent—turns a single interaction into a coordinated, cross-channel experience that feels like a natural extension of the brand

### Accessibility Is Non-Negotiable

Credible try-on must perform consistently across diverse customers—not just the narrow demographic often used in technology testing. This isn't just ethical; it's essential to building confidence and reducing returns across your entire customer base.

Requirements for inclusive virtual try-on:

- **For beauty**: Rigorous shade and undertone modeling that accounts for different lighting and camera conditions, with calibration that avoids over-smoothing and preserves real texture
- **For eyewear and jewelry**: Precise scaling, fit guidance, and clear, non-technical explanations that help users understand how size and proportion translate from screen to real life
- **Multiple modes**: Camera-on self try-on, try-on on models with similar features, and accurate static previews for customers who prefer not to use live AR
- **Navigation support**: Ensure each mode can be navigated with keyboard and screen-reader support
- **User control**: Transparent controls to adjust lighting, skin tone matching, and fit markers, so users can fine-tune results to their context rather than accept a one-size-fits-all output

Investing here is both the right thing to do and good business: it reduces friction, broadens reach, and builds trust with audiences often underserved by generic experiences, while improving confidence in purchase decisions and, ultimately, lowering return rates.

## Measuring Success Beyond Conversion Rates

Evaluating the success of virtual try-on requires a connected narrative that ties commercial impact to experience quality and brand equity. In luxury, it can be the right trade-off to accept slightly lower short-term conversions if the experience feels more authentically premium and strengthens brand equity over time. This is the most important tension to understand: **brand perception and conversion aren't always aligned, and in luxury, brand perception often matters more.**

Start with the financials to confirm tangible value—conversion rate lift, return-rate reduction, and AOV growth—then use experience signals to explain why those outcomes occur. Post-interaction surveys should assess whether the experience elevated brand perception, not just whether it drove conversions. 

Track the metrics that reveal whether virtual try-on is reinforcing your luxury positioning:

- **Engagement depth**: Time spent in try-on, number of products and variants explored, completion of key actions such as saving to wishlist or adding to cart
- **Downstream behaviors**: Do users who engage with try-on have higher LTV, lower return rates, or greater repeat purchase frequency?
- **Cross-channel continuity**: In-store visits that reference an at-home try-on shortlist demonstrate how digital confidence translates into physical conversion
- **Brand outcomes**: Pair pre/post brand perception scores (quality, innovation, trust) with sentiment from social and customer service that explicitly references try-on
- **Material authenticity correlation**: Analyze whether improvements in perceived fit and material authenticity correlate with reduced returns and higher AOV

When these layers align—stronger brand scores, deeper engagement, and better unit economics—you have the connective tissue that shows virtual try-on is not merely lifting conversion, but reinforcing the luxury positioning that sustains pricing power and long-term value.

## Technology Should Serve the Brand, Not Define It

AI-driven virtual try-on isn't a trendy feature—it's a tool to increase customer confidence, reduce returns, and strengthen brand positioning. But only when implemented with the same attention to detail that luxury brands apply to every other customer touchpoint. The brands succeeding with virtual try-on share a common principle: technology should serve the brand experience, never define it. The technology should feel invisible, allowing the product and brand to remain center stage. 

The gap between leaders and followers will only widen as technology continues to evolve. Brands that invest strategically now in high-quality implementations will establish competitive advantages that become increasingly difficult to replicate. But implementation quality matters far more than speed. A poorly executed virtual try-on can do more harm than not having one at all. 

The key is finding the right balance between innovation and brand integrity, between cutting-edge technology and timeless luxury values. When you strike that balance, virtual try-on becomes more than a feature—it becomes an extension of what makes your brand special. 

**Need help implementing a virtual try-on experience that actually serves your brand?** Let's discuss creating an experience that maintains your luxury positioning while delivering tangible business results.

Why Luxury Brands Are Betting on AI-Powered Virtual Try-On

In e-commerce, a smooth and intuitive shopping experience can make all the difference—helping customers find what they need faster, increasing the chances they'll complete a purchase, and increasing their basket size.

While Shopify offers several features to enhance product presentation and purchasing flows, many merchants struggle with how to effectively display related products—such as variations, complementary items, and customizable sets—while keeping the their online store’s shopping experience intuitive.

In this article, we're going to showcase three different ways to present related products in a Shopify store and give you a clear framework for understanding what tool to use depending on your specific scenario.

Let's take a look!

## LAKE's Product Presentation Challenge: A Case Study

While working with premium sleepwear brand LAKE, we encountered multiple product presentation challenges that illustrated the importance of selecting the right tool for the job. LAKE's diverse product catalog—featuring style variations, complementary products, and customizable bundles—called for a deep understanding of product relationships before implementing technical solutions.

By analyzing how LAKE's products related to one another and how customers interacted with them, we were able to present an approach that addressed their specific business objectives. Specifically, LAKE wanted to explore different business scenarios:

* Enhance the product browsing experience by showing style variations (like sleeve length) on product pages.
* Implement more effective product recommendations using straightforward solutions.
* Create a customizable experience for bundled items that were being treated as single products.

After a thorough analysis of their customer journey, we discovered that no single Shopify app would address all these needs. Instead, we explored a hybrid approach based on the different types of product relationships.

## Three Solutions for Product Cross-Selling

Based on LAKE's needs and the different types of product relationships, we proposed three distinct solutions for cross-selling:

### 1. Combined Listings for Style Variations

For closely related style variations (like the same pajama top in short-sleeve and long-sleeve versions), we recommended Shopify Combined Listings. This feature (available exclusively on Shopify Plus and enterprise-level plans) allows merchants to display multiple related products as variants within a single listing while keeping them as separate products in the backend.

This means each product can have its own:

* Product title, product description, and metadata
* Unique image gallery
* Independent product URL for better SEO

Why it works for style variations:

* Customers can easily toggle between short-sleeve and long-sleeve options without leaving the page
* The variations are genuinely different versions of the same core product (same fabric, design, and general style)
* It provides a seamless browsing experience while keeping separate product URLs for better SEO and inventory management

<figure>
	<img src="https://cdn.sanity.io/images/kqgmf0g2/production/0e7ef79ee680f895fdcfa1e7762645e7c17c4ed7-1600x1052.jpg?fit=clip">
  <figcaption>A prototype for LAKE, showcasing a men's pajama set with two style variations.</figcaption>
</figure>

Another use case is using Combined Listings to display separates (i.e. top, bottom) that are often bought together.

<figure>
	<img src="https://cdn.sanity.io/images/kqgmf0g2/production/13771e641a9ab5940cc706364b59b131a59c4491-1600x1052.jpg?fit=clip">
  <figcaption>An example of how separates can be shown usinf Combined Listing.</figcaption>
</figure>


### 2. Quick Add to Cart for Complementary Products

One of the options LAKE asked us to explore was a flexible way to display matching products on the product page. For example, a women's pajama set and its matching version for children are perfect for a "mommy and me" moment. While these items are designed to complement each other, they are separate products targeting different users, not variations of the same item.

In this case, a Quick Add to Cart functionality works best:

* A women's pajama set and a child's pajama set are separate products rather than variations of the same core item, unlike short-sleeve and long-sleeve versions of a t-shirt
* A Quick Add button allows customers to add the matching children's set to their cart instantly, making it easier to purchase both and increasing the overall AOV
* Each pajama set retains its own product description, sizing chart, and imagery, ensuring clear product details and avoiding confusion

<figure>
	<img src="https://cdn.sanity.io/images/kqgmf0g2/production/ba256f0f864d3111809f4a6648b5e700e867cee6-1600x1052.jpg?fit=clip">
  <figcaption>Another prototype for LAKE, to explore the Quick Add to Cart solution.</figcaption>
</figure>

### 3. Bundling Apps for Customizable Sets

When customers need to personalize their purchases beyond what Combined Listings can offer, bundling apps from the Shopify App Store provide the necessary flexibility while still creating a cohesive shopping experience. These specialized apps excel in scenarios where:

* Customers need to select products in different sizes for related items (like tops and bottoms)
* Merchants want to offer mix-and-match sets with special bundle pricing as an upsell
* There's a need to maintain accurate inventory tracking for each item

For example, HoneyLove sells women's underwear, providing a sophisticated shopping experience that allows customers to select different sizes for tops and bottoms without navigating between product pages, streamlining the path to checkout.

<figure>
	<img src="https://cdn.sanity.io/images/kqgmf0g2/production/c31fe6803eda7868b77853a1f1dd8d8a0f523ea2-3200x2104.jpg?fit=clip">
  <figcaption>Honeylove showcases an activewear bundle with coordinated separates, allowing customers to mix and match pieces.</figcaption>
</figure>


## Implementing the Right UX for Your Cross-Selling Needs

As we've seen, these approaches aren't mutually exclusive—they can be combined strategically based on your specific business goals and product relationships.

To help streamline the decision between different cross-selling experiences, we suggest looking at your product relationships through two key dimensions:

* **Variation Complexity:** How different are the products? This ranges from simple variations like color and size to entirely separate but complementary products.
* **Purchase Intention:** How likely are customers to want these items together? This can range from items that are almost always purchased together to those that are only occasionally complementary.

By understanding where your products fall on these dimensions, you can choose the right tools to create an intuitive shopping experience that aligns with your business goals.

When you take time to analyze how your products relate to one another, you create shopping experiences that improve customer experience and increase transaction value, boosting sales and setting your brand apart from competitors.


The Nebulab

This Is Why Your E-Commerce CRO Program Isn't Working

Why Luxury Brands Are Betting on AI-Powered Virtual Try-On

Related Products in Shopify: Picking The Right Tool for the Job