Fashion6 min read
Selling bikes and e-bikes online in 3D: frame size, accessories, configurators
What a 3D viewer and a configurator actually answer when someone is buying a bike online, and what they still cannot tell a rider.
By CharpstAR · 22 Sept 2026
A bike is one of the hardest things to sell from a photo. It is expensive, it comes in sizes that are described in numbers nobody can picture, half of what a buyer pays for is bolted on afterwards, and the colour matters more than any retailer wants to admit. Bike shops have known this forever, which is why the traditional answer is "come in and we will put you on one."
Online, that option does not exist. So the question for anyone selling bikes or e-bikes on a website is narrower than "should we do 3D." It is: which of the questions that used to be answered in the shop can a 3D model on a product page actually answer, and which ones cannot be answered that way at all.
The four questions a bike buyer has
Strip out the browsing and it comes down to four things.
Will it fit me. Frame size, standover height, reach, seat height range. This is the question that keeps people out of the checkout.
What does it actually look like. Colour under normal light, the shape of the frame from the side and from three quarters, how thick the downtube is on an e-bike once the battery is in it, whether the welds are tidy.
What comes with it, and what do I add. Racks, mudguards, lights, panniers, a child seat, a bigger battery. On cargo bikes this is most of the purchase.
Where will it live. Cargo bikes and longtails do not fit in a normal hallway or a normal lift. People discover this after delivery.
A 3D model on the page answers the second, third and fourth question well. It answers the first one only partly, and honesty about that split is what separates a useful project from a disappointing one.
Frame size is the part 3D does not solve
Fit on a bike is a relationship between the frame and the rider's body: inseam, torso, arm length, flexibility, how they ride. No 3D viewer measures a person, and there is no phone-based body scan today that gets a bike fit right. Anyone selling you a bike configurator on the promise that it ends size returns is overselling.
What 3D does do for sizing is make the geometry legible. A frame you can rotate, with standover height and seat height range drawn on it as dimension lines, is easier to reason about than a geometry table with eighteen rows of millimetres. Put a human silhouette of a stated height next to it and the comparison becomes something a shopper can actually do. That is a real improvement over a spec table, and it is not the same as a fitting.
The rest of sizing is handled by boring, non-3D things: a size chart keyed to rider height, clear inseam guidance, and a returns policy that says what happens if the size is wrong. Do those first.
Where AR earns its place: the bike in your actual space
The fourth question is the one AR answers better than anything else. Both mobile platforms will place a correctly scaled model in a room straight from the browser: Android does it through Scene Viewer, where one unit in the glTF file equals one metre so the model appears at true size, and iOS does it through AR Quick Look in Safari. No app, no download.
For a standard road or hybrid bike this is mildly useful. For anything long, it is the whole ballgame. A cargo bike is around two and a half metres of vehicle. People buy one and then find out it does not turn in their hallway, does not fit the bike shed, will not go into the lift. Letting someone stand in their own hallway and place the bike there before ordering removes an expensive category of return, the kind where the item comes back damaged in a box that cost sixty euros to ship.
Accessories are the reason to build a configurator
We built a configurator for Livelo, who sell cargo bikes, and the thing that made it worth building was not the frame colour. It was that almost nothing about a cargo bike is a single decision. Frame and colour, then the box or the flatbed, then the rain cover, then child seats, then lighting, then a battery size that depends on how heavily the owner plans to load it.
Sold as a product page with a long options table, that is a form. Sold as a configurator, it is a bike on screen that visibly gains a box when you pick the box, gains a cover over the box, and grows two seats inside it when you add children. The accessory stops being a line on an invoice and becomes part of the picture of the thing the buyer is about to own.
We are not going to attach a number to that, because we have not measured it in a way that would survive scrutiny and there is no honest published figure for configurator attach rates. What can be said about the mechanism is that the accessory is shown in context, on the product it belongs to, at the moment the decision is still open. Every one of those three is a known weakness of the related-products row at the bottom of the page.
You can try the same pattern on a different product on our configurator page, where the modules and fabrics of a sofa behave the same way.
What it is reasonable to expect from the numbers
The published evidence on 3D and AR in e-commerce is thinner than the marketing suggests, and none of it is bike-specific. The figure most often quoted is Shopify's, reported in 2020: interactions with products carrying 3D or AR content converted 94 percent more often than the same kind of products without. That is a comparison across merchant data rather than a controlled test, and merchants tend to model their better products first, so read it as a direction rather than a forecast.
The more concrete numbers Shopify published come from Rebecca Minkoff and Gunner Kennels: 44 percent more likely to add to cart after interacting in 3D, 27 percent more likely to place an order, a 65 percent purchase lift on AR-enabled pages, and a 5 percent reduction in return rate at Gunner Kennels. Different categories, different price points, but they give you the shape of the effect.
On returns, the money is real. The National Retail Federation put United States returns at 743 billion dollars in 2023, 14.5 percent of sales and 17.6 percent online. Bikes are heavy, so each one that comes back costs several times what a returned t-shirt costs.
A sensible order to do this in
| Step | What you build | Why first |
|---|---|---|
| 1 | 3D model of your two or three best-selling frames, with dimension lines | Covers most of your traffic, proves the format on your own analytics |
| 2 | AR button on those pages | Answers the "where does it live" question, which is where the expensive returns come from |
| 3 | Accessory configurator on the cargo or longtail range | Highest number of combinations, highest basket, most to gain |
| 4 | Reuse the models in ads, email and print | The model is already paid for by this point |
Do not start by modelling the entire catalogue. Bikes share frames, wheels and components across a range, so once the first few models exist the rest are much cheaper, but there is no reason to find that out on forty products instead of three.
Try it on one bike
If you sell bikes, the fastest way to decide whether any of this applies to your shop is to put one model live and watch what your own customers do with it. We will build a 3D model of one of your products for free so you can do exactly that, and the model is yours to keep. Send us one product. What we build beyond that, and what it costs, is on solutions and prices.
Sources
- Google, Scene Viewer documentation.
- Apple, AR Quick Look.
- Shopify, merchant data on 3D and AR content, September 2020.
- Shopify, AR shopping: Rebecca Minkoff and Gunner Kennels results.
- National Retail Federation and Appriss Retail, 2023 Consumer Returns in the Retail Industry.
