Driver one: native or web
This is the biggest fork. A WebAR experience runs from a link in the phone browser with no install, which makes it dramatically cheaper to build and distribute, and it is usually the right answer for a campaign, a product page or a one-off activation. A native ARKit or ARCore app costs more but gives better tracking, offline use and device features. Choosing native when WebAR would do is the most common way to double a budget for no gain.
Driver two: how many 3D assets
For product AR this is usually the largest single line, and it scales with your catalogue rather than your app. Ten accurate, optimised, real-scale models is a very different project from two hundred. This is why we tell retail clients to start with their ten best sellers: it proves the conversion lift and establishes the pipeline at a fraction of the cost.
Driver three: what the AR has to understand
Placing an object on a detected floor is well-trodden. Image and marker tracking is straightforward, and so is face tracking. It gets expensive when you need persistent anchors across sessions, multi-user shared AR, precise measurement, or occlusion behind real objects. Each of those is a real engineering problem rather than a configuration option.
Driver four: integration
A standalone demo is cheap. An AR viewer wired into a live product catalogue, with inventory, pricing, variant selection and analytics flowing back into your systems, is a software integration project that happens to have AR in it. Be clear which one you are buying.
Realistic ranges
A focused WebAR demo or single-product viewer starts at $999 in our Spark package. A full AR application with a real catalogue, custom interactions and analytics starts at $2,999 under Launch. Enterprise AR, meaning industrial work instructions, multi-user or deep systems integration, is quoted individually. The cost calculator gives a ballpark on your own spec.
Where teams overspend
Building native when WebAR would reach more people. Modelling the whole catalogue before proving anyone uses it. And commissioning photoreal assets that a phone cannot render smoothly anyway. All three are avoidable in the scoping conversation. Try our live WebAR demos before you decide.
Related: AR development · How AR furniture apps work · Cost calculator
Quick answers
How much does an AR app cost in 2026?
A focused WebAR demo or single-product viewer starts around $999. A full AR application with a real catalogue, custom interactions and analytics starts around $2,999. Enterprise AR with multi-user or deep systems integration is quoted individually.
Is WebAR cheaper than a native AR app?
Usually significantly, and it also removes the app-install barrier. Native ARKit or ARCore gives better tracking, offline use and device access, which matters for some products. Choosing native when WebAR would do is the most common way to double a budget unnecessarily.
What makes AR projects expensive?
Four things: choosing native over web, the number of 3D assets you need, advanced capabilities like persistent anchors, shared multi-user AR or occlusion, and integration with live catalogue, pricing and inventory systems.
How many 3D models should we start with?
Ten best sellers. That proves the conversion lift and establishes the asset pipeline for a fraction of the cost of modelling a full catalogue before you know anyone uses it.
Related guide
ARKit vs ARCore vs WebXR: which to build on →Costing an AR project?
Send us what you sell and we will come back with an approach, an asset plan and a fixed quote within 48 hours.