Furniture is the category where augmented reality makes the most obvious difference. A sofa is expensive, difficult to return, and almost impossible to judge from a photograph. Customers are not asking "what colour is it" — they are asking "will it fit, and will it look right in my room". Photography cannot answer that. AR can.
This guide covers what actually works, what it costs, and how to implement it without a development team.
The problem AR solves in furniture
Every furniture retailer deals with the same three recurring issues:
Size misjudgement. Customers read the dimensions, picture the piece in their head, and get it wrong. A 220cm sofa sounds fine until it arrives and dominates the room. This is the single largest driver of furniture returns, and returns in this category are expensive — you are paying for two-way freight on a bulky item.
Style uncertainty. A chair photographed in a bright minimalist studio looks different in a small, warmly lit flat. Customers hesitate because they cannot picture it against their own floor, walls and existing furniture.
Long consideration cycles. Furniture purchases involve deliberation, comparison, and often a second person who needs convincing. Anything that shortens that cycle or helps the customer make the case at home has direct commercial value.
What an AR furniture experience looks like
The mechanic is simple. The customer opens a link or taps a button on the product page. Their phone camera activates, they point it at the floor, and the piece appears at true size in their actual room. They can walk around it, view it from the sofa they already own, and check clearances against a doorway.
On iPhone this runs through Apple Quick Look; on Android through Google Scene Viewer. Both are built into the operating system, so there is no app to download. That matters enormously — a customer who is asked to install an app to view a product will simply not do it.
The decisive moment is not when the customer sees the product — it is when they see it next to something they already own. A dining table placed beside their existing chairs, or a bookcase against the wall it will actually stand against. That comparison is what turns hesitation into a decision.
Three ways to deploy it
1. On the product page
An embedded 3D viewer with a "View in AR" button, sitting alongside the photography. This is the highest-traffic placement and the one most retailers start with. It works on Shopify, WooCommerce, Wix, Webflow and custom platforms through a simple iframe embed.
Practical note: place it after the first image gallery, not before. Customers want the photography first; AR is the confirmation step.
2. QR codes in the showroom and in print
Showrooms cannot display every colourway, fabric or size. A QR code on the display card lets a customer scan and see the variant that is not physically present — in their own home, immediately.
The same applies to catalogues, direct mail and delivery inserts. A code on the packaging of a small item can drive the customer toward the matching larger piece.
3. Direct links in sales conversations
For retailers with a consultative sales process, sending an AR link by WhatsApp or email during the conversation is the highest-converting use of all. The customer is already engaged, and you are handing them the answer to their main objection.
This is also the easiest to trial — it requires no website changes at all.
Getting 3D models of your furniture
This is the part retailers worry about most, and it is less difficult than it used to be.
You may already have the models. If pieces were designed in CAD software, the geometry exists. A 3D generalist can convert and texture a CAD file far faster than modelling from scratch — often a few hours per piece rather than a few days.
Scanning is now viable. Free phone apps can produce a usable GLB from a thirty-second walk around a physical piece. Quality varies with lighting and surface type — matte fabrics scan well, glossy and transparent surfaces poorly. Good enough for a catalogue tail, not for hero products.
Commissioning is affordable per unit. A clean, optimised model from a freelance 3D artist typically costs between fifty and three hundred dollars depending on complexity. For a piece you will sell for several years, that amortises to almost nothing.
A sensible sequencing: commission proper models for your top twenty sellers, scan the mid-tail yourself, and leave the long tail without AR initially. The full guide to producing AR models covers each method in detail.
Technical requirements that matter
| Specification | Target | Why |
|---|---|---|
| File format | GLB | Works on both iOS and Android from a single file |
| File size | Under 10MB | Customers are often on mobile data |
| Polygon count | Under 100k triangles | Mid-range Android phones struggle above this |
| Texture resolution | 2048×2048 max | Larger sizes inflate file size without visible benefit |
| Scale units | Metres | A model built in centimetres appears 100× too large |
| Origin point | Base centre | Ensures the piece sits flat on the detected floor |
The scale and origin points are where most first attempts fail. If your sofa appears the size of a building or floats above the floor, one of those two is wrong.
What it costs to start
A realistic budget for a retailer testing AR on their top products:
- Platform: free to start, around five dollars a month for unlimited products on a tool like ARView. Dedicated enterprise AR platforms run considerably higher.
- 3D models: zero if scanning yourself, or roughly fifty to three hundred dollars per commissioned model.
- Implementation: an iframe embed pasted into your product template. Ten minutes, no developer required for most platforms.
Testing five products properly is achievable for a few hundred dollars and an afternoon of setup. That is a small enough commitment to justify running the experiment before making a broader case internally.
How to measure whether it worked
Run AR on a subset of products and compare against the rest of your catalogue over the same period. The metrics worth watching:
- Return rate on AR-enabled products versus your site average. This is where furniture AR earns its keep.
- Conversion rate on those product pages.
- AR engagement rate — what proportion of mobile visitors actually tap through to AR. If this is very low, the button placement is wrong.
- Time on page as a secondary signal of engagement.
Give it at least thirty days and enough traffic to be meaningful. Furniture purchase cycles are long, so a two-week test will not tell you much.
Common mistakes
Starting with the whole catalogue. Modelling four hundred products before validating the concept is how AR projects stall. Start with five.
Hiding the AR button. If it sits below the fold under a specifications table, nobody finds it. Put it with the imagery.
Neglecting model quality on hero products. A poorly optimised or inaccurately scaled model actively damages confidence. It is better to have five excellent AR products than fifty mediocre ones.
Requiring an app. Any solution that asks the customer to download something will see single-digit engagement. Browser-based AR is not a compromise — it is the only version that works at scale.
Where to begin
Pick your three highest-return products. Get 3D models made for them — scan them yourself if budget is tight. Upload them, embed the viewer on those three product pages, and run it for thirty days against a control group.
The data from that test will tell you whether to expand, and give you the internal case to do it. Upload your first model free to see the customer-facing experience before committing to anything.