How AR Fitting Rooms Are Actually Changing What Shoppers Do at the Shelf
Last October, I was working with a mid-sized apparel brand in Atlanta — around 340 stores across the Southeast — that had just rolled out Snap's Shopping AR Lens across twelve pilot locations. The brief was simple: reduce return rates on jeans, which were running at 34% nationally for them. Three months in, those pilot stores clocked a 19% drop in returns on the denim category specifically. That's not a rounding error. That's a logistics and margin story with real teeth.
I've been in retail tech implementation for about nine years, and the pace of actual AR adoption on the floor has surprised me over the last eighteen months.
What "Try Before You Buy" Looks Like When It's Not a Gimmick
The phrase has been abused. For years, "virtual try-on" meant a janky overlay that made you look like you were wearing a cardboard cutout of a shirt. What changed is the underlying depth-mapping and body-tracking accuracy. Tools like Zeekit (now owned by Walmart), Wanna Kicks for footwear, and the newer builds inside Shopify's AR product viewer are actually tracking fabric drape and fit contour in ways that don't feel like a tech demo anymore.
The Atlanta deployment used a combination of in-store iPad kiosks running a custom Unity build alongside a QR-triggered mobile experience. Customers could scan a product tag, see the item on a photorealistic body scan that matched their inputted measurements, and then compare two colorways side by side. Average dwell time at the kiosk: 4.2 minutes. That's meaningful in a retail environment where the average shopper gives a fixture about 8 seconds of attention.
The Metric Retailers Are Measuring Wrong
Here's the thing a lot of brands get wrong: they track AR session starts and call it engagement. That's like counting how many people picked up a shirt and put it back down.
The number that matters is conversion lift in the SKUs that have AR experience attached versus those that don't, measured over a minimum 90-day window. In a sporting goods client I worked with in March 2024 outside of Denver, AR-enabled products showed a 27% higher add-to-cart rate than identical products without the overlay — same price, same placement, same promotional support. The feature was the variable.
Don't get distracted by session counts. They'll make you feel good and tell you almost nothing useful.
I Used to Think In-Store AR Was the Priority. I Was Wrong.
For a long time I pushed clients hard toward in-store kiosk deployments. My reasoning was that the in-store environment was controlled — you could manage lighting, device hardware, and the customer's attention span. Mobile was too fragmented, too dependent on camera quality, too many variables.
I changed my mind around mid-2023 after seeing post-purchase survey data from a home furnishings retailer using IKEA's Place app methodology as a reference model. Seventy-one percent of customers who reported using any AR feature before a furniture purchase had used it at home, not in-store. They were scanning their actual room. The context that drove confidence wasn't the store — it was their own living room at 9pm on a Thursday. That insight flipped my entire implementation priority order.
Now I recommend mobile-first AR experiences with in-store as the reinforcement layer, not the anchor.
Where the Friction Still Lives
It'd be dishonest to pretend this all works smoothly. The deployment gaps I still run into constantly:
- Product catalog readiness. Most mid-market retailers have fewer than 15% of their SKUs with 3D assets. Without that, you can't build an AR experience. The asset creation cost is real — roughly $40–$120 per SKU for a decent 3D model depending on complexity.
- Staff training lag. Customers walk up to a kiosk and associates can't explain it. That kills adoption faster than any technical bug.
- Android fragmentation. ARCore support is inconsistent across budget Android devices, and a significant portion of lower-income shoppers are on hardware that can't run the experience reliably. That's an equity gap worth talking about.
The asset problem is the one most brands underestimate at the planning stage. I've watched two rollouts stall for six months not because of technology but because the 3D catalog wasn't ready.
What's Actually Coming in the Next 18 Months
Generative AI is starting to merge with AR product visualization in ways that are genuinely interesting. There are early builds — companies like Hexa and Vertebrae are doing active work here — that can auto-generate 3D assets from a 2D product image in under four minutes with accuracy that's getting close to usable. When that pipeline matures, the catalog readiness problem starts to dissolve, and the economics of AR deployment change completely for smaller brands.
The brands that are quietly building their 3D asset libraries right now are going to be in a very different position in 2026.