Augmented reality for a small shop: what a few hundred euros actually buys

Articles about AR for small business tend to describe what enterprise retailers do and then suggest you do a smaller version of it. That is not useful if you sell ten products and your marketing budget for the quarter is a few hundred euros. So this is a budget-first piece: what that money buys, what it does not, and how to tell whether you should spend it at all.

The good news about platforms

AR on the web is free to deliver. You do not need an app, a licence, or a partnership.

On Android, Google's Scene Viewer opens a 3D model in the camera from an ordinary web link, on any ARCore device running Android 7.0 or later. On iPhone and iPad, AR Quick Look does the same in Safari from iOS 12 onwards. Both are built into the operating system. The `` web component that drives most implementations is open source and free.

So the entire cost of AR for a small shop is the cost of the 3D models, plus an hour of someone's time to add a snippet to a product template. That is the correct way to frame the budget.

What a model costs, honestly

For a simple hard good with clean geometry, a single production-quality model is a low hundreds of euros as a one-off, or a monthly fee if you want it maintained, hosted and optimised. Our own pricing starts at 10 dollars per product per month on Basic, on a twelve-month term with a minimum of 100 variants, and 20 dollars per model per month on the Value tier with a twenty-four month term. Those minimums exist because the work is in the pipeline, not the individual file, which is the honest reason small volumes cost more per unit everywhere in this industry.

Shopify's own guide puts custom AR app development at 30,000 to 50,000 dollars for basic functionality and up to 300,000 for complex work, and prebuilt AR apps at around 39 dollars a month. If anyone quotes you the first range for putting a sofa on a product page, walk away. That is app development pricing, and you do not need an app.

The realistic plan for a ten-product shop

Do not model all ten. Model one or two. Pick the products where size or material is the reason people hesitate, and where the order value justifies the spend. A 900 euro cabinet earns a model. A 25 euro bookend does not.

Start with a free one. We build one model at no cost, and we are not the only ones who do something similar. Use it on your best-selling product, leave it up for a month, and compare conversion and returns against a comparable product without it. That test costs you nothing but attention and it will tell you more than any industry statistic.

Do the cheap page work first. Accurate dimensions, more photographs in real rooms with a recognisable object for scale, an honest description of the material, and a clear returns policy. All of that is free and it addresses the same shopper anxiety AR addresses.

Skip the DIY scanning unless you enjoy it. Phone-based photogrammetry produces a usable rough mesh in an afternoon. Getting it to look like your product on a white page takes an artist. If you have the skills and the time, the tools are genuinely accessible now. If you do not, the result will look worse than your photographs, and a bad model actively costs you sales.

When the answer is to wait

Be honest with yourself about three things.

If your products are small, cheap, or bought on impulse, AR does not answer any question your shopper is asking. If your traffic is low enough that you cannot tell the difference between a good week and a bad one, you cannot measure the effect and you will be spending on faith. And if your product page has an unclear shipping cost or a forced account at checkout, fix those first. Baymard's meta-analysis of 50 studies puts average cart abandonment at 70.22 percent, with extra costs at checkout cited by 40 percent and forced account creation by 18 percent. No 3D model will beat fixing those.

When the answer is yes

If you sell something people place in a room, wear on their face, or need to picture at true size, and your average order is high enough that one extra sale a month covers the cost, the case is straightforward.

That is why the categories that adopt this first are furniture, lighting, heating and eyewear. Contura sells wood-burning stoves, which almost nobody has bought before and which are hard to imagine at scale. MELIMELI sells sofas in more fabric combinations than anyone could photograph. Those problems exist at ten products just as much as at a thousand.

What good looks like on a small site

One thing matters more than the rest: speed. Google recommends models under 10 MB for Scene Viewer, ideally 30,000 to 50,000 triangles, and Shopify suggests around 4 MB as the target for a product page. A heavy viewer that slows your page will cost you more traffic than the AR earns you. Ask whoever builds your model what the file size is before you accept it.

Beyond that, put the 3D viewer in the main image area rather than hiding it in a tab, label the AR button plainly, and make sure the model matches the product you actually ship.

The cheapest next step

Pick your best-selling product and send it to us for a free model. Put it live, watch it for a month, and decide with your own numbers. If it works, our pricing and what we build are there. If it does not, you have lost nothing and you know something real about your catalogue.

Sources

• Google, Scene Viewer developer documentation.
• Apple, AR Quick Look.
• Shopify, AR shopping, including AR app cost ranges.
• Shopify, 3D e-commerce, including model size guidance.
• Baymard Institute, cart abandonment rate statistics.