The sofa in your room before you buy it

Nobody buys a sofa without measuring first

Olmo is a brand I invented to work on a real problem: selling expensive furniture from a phone. A Madrid design furniture brand, warm and made of natural materials, with a physical store and an app that is not just the website squeezed down.

Furniture has something fashion does not: you cannot try it on. You imagine it. And what people imagine is almost always the same: whether it fits the space, whether it goes through the door, whether the fabric matches the photo and what it will cost in the end with delivery and assembly.

I worked the whole project end to end, in the same order I follow with a client: understand the market first, then decide, and only then draw.

What the data says before drawing anything

I started with the market and with how people buy furniture online: sector data for Spain and the usability research from Baymard Institute and Nielsen Norman Group, which test real users on furniture stores.

The finding that changed the project was about augmented reality. Everyone uses it as a hook, yet in Baymard's tests 87% of users who saw the "view in your room" button chose not to use it. Not because they don't want to see it at home, but because the experiences they know work badly: imprecise controls, models that resize by accident and colours that don't match the photo.

Ten stores, and half of them without an app

I compared the stores competing in Spain for the same customer, plus two international references, on the same criteria: whether they have an app, what they do with AR, how they handle delivery and assembly and which payment methods they offer.

The finding is a gap: Kave Home, Sklum and Maisons du Monde, the three affordable design brands, have no shopping app. On mobile they only sell through the website. The ones that do have an app all sit at 4.7 stars, so the bar is not having an app, it's making it worth the download.

Eight decisions, each with its data point

The research and the benchmark produced eight opportunities. None is an opinion about taste: each points to a data point or a specific gap in the comparison, and each became a design decision.

Five tabs and three journeys

The app is organised in five tabs: Home, Explore, Spaces, Favourites and Profile. Search, profile and menu always sit top right on every screen, and the cart lives inside the menu with its counter, so the header doesn't grow with every new feature.

I mapped three journeys with their alternative branches: discovering and buying a sofa, first use and photo search. The branch I cared about most is in the first one: if the sofa doesn't fit in AR, the app takes you back to the list with the size filter already applied instead of leaving you at a dead end.

Grey, and already clickable

The nine wireframes cover the whole purchase flow, from splash to confirmation. They are grey and unbranded on purpose: at this stage the discussion is what is on each screen and in what order, not the colour of the button.

Low fidelity as they are, they were already wired as a prototype in Figma, and each screen carried a note with its planned animation. That way motion was decided with the flow in mind rather than added at the end as decoration.

Real materials, turned into tokens

The palette comes from the product itself: linen sand, bouclé terracotta, oak and olive. They are the colours of the fabrics the brand sells, so the interface never competes with the sofa photo; it frames it.

Terracotta is reserved for what you can act on. Olive is secondary and belongs to AR. The background is cream, not white, because pure white makes product photos look cut out.

Product photography first, interface behind it

The product and room photos are AI-generated in Higgsfield, making sure every sofa shares the same light, angle and floor. That is what makes a catalogue look like one brand and not a collage of suppliers.

The Duna sofa page has its five fabrics photographed in the same framing, so changing fabric changes the whole photo, not a colour filter on top. The monthly price sits under the price, and dimensions with door clearance and three free swatches are one tap away.

Real scale, with the centimetres in view

The first version of the viewer was the one everybody makes: the sofa on top of a photo of the room. When I tried it I found the same problem Baymard describes: the sofa looked small, couldn't be moved and there was no way to tell whether it fit. So I rebuilt it as a 3D viewer.

You look around the room in 360° by dragging. You tap the sofa and take it anywhere on the floor, rotate it with two fingers, and it never changes size: it always measures its 180, 220 or 260 cm. While you move it you see the centimetres left to each wall, and if it hits the plant or the lamp, it says so in red.

There is a floor plan view for precise placement, a 1.70 m person for scale and two ways to use your own home: enter your living room's width and length, or upload a 360° photo taken with your phone from the middle of the room.

Usable before it exists

What you see in the video is the clickable prototype running, not an edit: every tap is marked with a circle. I built two versions. In Figma the screens are wired with Smart Animate, Push, Move-in and Dissolve. In parallel I built an HTML, JavaScript and Three.js version with the real animations, the side menu, a cart that recalculates and the 3D AR viewer.

The coded version has a product reason: an AR viewer can't be prototyped in Figma. To know whether moving a sofa with your finger makes sense, you have to be able to move it.

How it looks off the computer screen

A mobile design is judged on a phone. I generated the scenes in Higgsfield with a green screen on the phone, located the rectangle and placed the real Figma screens inside, so what you see on the iPhone is the exact design and can be updated if it changes.

A product backed by data, ready to test

Olmo is a concept: it has no users yet, so I don't show conversion figures I haven't measured. What it does have is every decision tied to a data point or a market gap, and a prototype that can be used to test them.

The next step is the one I follow on any product: usability testing with five or six people who bought a sofa online in the last year, with three tasks —find a sofa that fits a 2.40 m space, change its fabric and place it in AR— and then iterate. And in the next version, reviews with photos of real homes and end-to-end swatch ordering.