A buyer's guide to 3D and AR for furniture retailers

How a furniture retailer should choose a 3D and AR supplier: the questions to ask, the traps in the pricing, and what a good first project looks like.

By CharpstAR · 22 Sept 2026

SEP2026

Augmented Reality: Revolutionizing Online Furniture Shopping

Every furniture retailer has now had the demo. A sofa spins on a laptop, someone holds up a phone and the sofa appears on the meeting-room floor, and the question that follows is what it would cost to do that for the catalogue. The demo is the easy part. This guide is about what comes after: how to compare suppliers, what to ask, where the pricing hides its costs, and what a sensible first project looks like. It is written by a supplier, so read it as a checklist you can hold us to.

First, be clear what you are buying

There are four things a furniture retailer can buy under the heading "3D and AR", and they are priced differently.

The models. Someone has to build a 3D model of each product from your photos and drawings, with real materials. This is the labour, and it is where quality lives.
The viewer. The code that shows the model on your product page: rotate, zoom, dimensions, hotspots. Usually a script tag or a plugin.
The configurator. The viewer plus your product rules: fabrics, legs, modules, prices. This is a project, not a plugin.
AR delivery. The same model exported so a phone opens it in the shopper's room from a web link, without an app. On modern phones this is standard; if a supplier needs an app, ask why.

Some suppliers sell only software and expect you to bring models. Some sell only models. Some, including us, sell the whole chain as a monthly fee per product. None of these is wrong, but a software price with no models attached is not comparable to a full-service price, and that is the first trap.

The questions to ask

Who builds the models, and from what? Photos and drawings, a scan, or generated. Photogrammetry scans are fast and heavy, with baked lighting, and cannot make the fabric you have not manufactured yet. Generated models guess the back from the front. Hand-built models from drawings take about a working week each for furniture and are the only kind that can be reconfigured.

Show me the materials up close. Zoom into the fabric and the wood on the demo. Shoppers do this first. Repeating tiles and plastic-looking leather are the most common failure and they are visible in ten seconds.

Is the scale checked? Ask how dimensions are verified against the product sheet. A model 5 percent too big looks right and is wrong, and it breaks the one thing AR is for.

What happens when a fabric is discontinued? Catalogues change. Ask what a texture change costs, what a new module costs, and what happens to the models if you stop paying. A catalogue of models nobody maintains is wrong within a year.

Does AR need an app? It should not. iPhone and Android both open a 3D file into AR from a normal web link. An app is a choice a supplier made, and it costs you most of your shoppers.

Where are the files hosted, and can I have them? The models are your product data. Make sure they are yours, and that they are served from a CDN fast enough that the viewer is not the slowest thing on the page.

What do comparable retailers measure? Ask for numbers from a furniture retailer, not a platform average. The platform averages are real but come from self-selected shoppers.

The pricing traps

Model	Looks like	Watch for
Per model, one-off	Cheap per unit	Updates and variants cost extra; no one owns keeping it current
Software licence, bring your own models	Low monthly fee	The models are the expensive part and are not included
Monthly per product, all included	Higher per unit	Minimum volume and contract length; check what "included" covers
Enterprise custom	Flexible	Integration hours; the configurator rules are where the hours go

Our own model is the third row: a fixed monthly fee per product covering modelling, viewer, AR, hosting and yearly updates, with a minimum volume and a 12- or 24-month term. The numbers are on the prices page, and the reason for the shape is that it makes someone responsible for the catalogue staying right.

What the numbers say, and how to read them

The figures that will appear in every supplier's deck: Shopify's merchant data put conversion 94 percent higher on interactions with 3D or AR content; Rebecca Minkoff saw 65 percent more purchases among AR users; Gunner Kennels cut returns 5 percent and lifted conversion 40 percent. On the returns side, online furniture returns run about 22.7 percent, and about 58 percent of them are size and space mismatches, the exact thing AR placement addresses.

Read them as direction, not forecast. The conversion figures compare shoppers who chose to use 3D with those who did not, so intent is mixed in. The returns figures are the sounder argument for furniture specifically, because they describe a reason for returns that a model with dimensions removes.

What a good first project looks like

Twenty products, not two and not two thousand. Choose the ones with the highest return rate for size reasons or the most configuration questions to customer service; usually sofas, beds and large storage. Have them modelled properly, put the viewer above the fold with an AR button, and badge them on the category page so shoppers learn to look. Measure for a month: viewer use, AR use, conversion and returns on those SKUs against the month before. Then decide on the catalogue with your own numbers.

Two retailers we work with did versions of this. Sweef started with its modular sofa, the product with the hardest combination question, and built out from there. Contura started with the stoves where "will it suit the room" was blocking the dealer call, and built a configurator around exactly that question.

The offer

If you want to judge the models before anything else, the fairest test is one of your own products. We build it in 3D and AR for free, with dimensions and hotspots, and it is yours to keep whether or not you go further. Send us one product.