3D for e-commerce5 min read

Planning a 3D e-commerce budget by product count and phase

What a 3D model costs per product, how to size the budget on distinct geometries instead of SKUs, how to phase a large catalogue and what to cut first.

By CharpstAR · 21 Mar 2023 · Updated 7 Oct 2026

Part of 3D models for e-commerce: the complete guide

MAR2023

Planning a 3D e-commerce budget by product count and phase

At budget time 3D becomes a line item someone has to defend, and the defence depends on how the cost is counted: by distinct geometry instead of SKU, phased so the first year is affordable, and with a list ready of what to remove when finance says the number is too big.

How to count what needs modelling

Your catalogue has SKUs. A sofa in twelve fabrics and three sizes is 36 SKUs, and it's usually one to three base models with a material library applied on top. Each fabric variant costs a small fraction of a new model.

Before anyone quotes, count the distinct geometries and the variants per geometry. Distinct geometries are the physically different shapes. A table and the same table 20 cm longer are two geometries, and the second is cheap; a table and a sofa are two geometries that share nothing. Variants are the colours, finishes, fabrics and leg options, which are material swaps.

For a furniture catalogue of 2,000 SKUs, the distinct geometry count often lands between 150 and 400. The budget is built on that number.

What a model costs

A model is priced either once or monthly.

Paid once, you own the file. The cost scales with complexity, so a rigid box is cheap, a moulded chair costs more, and a configurable upholstered corner sofa with realistic fabric behaviour costs considerably more. You then carry the hosting, the format conversion and the maintenance yourself.

Paid monthly, the subscription covers modelling, hosting, conversion to GLB and USDZ, the viewer and ongoing updates. Ours works this way: Basic starts at $10 per product per month on a 12-month term with a minimum of 100 variants, Value starts at $20 per model per month on a 24-month term, and Enterprise is quoted against the catalogue.

Which is cheaper depends on whether your range changes. A stable catalogue with in-house 3D skills favours one-off ownership. A catalogue that turns over each season, and a team with no 3D capability, favours the subscription. File formats and viewers change every couple of years, and someone has to re-export when they do.

Costs outside the vendor invoice

Someone has to look at every model and approve or reject it. Budget a few minutes per product across one or two rounds. Across 300 models that's days of someone's time, and it needs a name against it.

Front-end work covers adding a viewer to the product page template, the poster image, the fallback for devices without AR and the analytics events. On a hosted platform this can be a couple of days. On a bespoke front end, budget one to three developer weeks, once.

Reference material can also cost. If your products aren't photographed from the back and underside and you have no CAD, somebody has to produce that. With good CAD that cost is small.

The savings that offset it

A 3D budget presented as a cost on its own gets compared with paid media. Present the offsets with it.

A model rendered at any angle, in any room set, in any colourway replaces part of a studio shoot. For catalogues that reshoot every season, this can cover the modelling. MELIMELI justified its 3D work on reusing the models across the shop and campaigns.

Support contacts fall where the page answers the question: "how wide is it", "does the cover come off", "will it fit under a 90 cm counter".

Returns are harder to estimate. The National Retail Federation and Appriss Retail put United States returns at 743 billion dollars in 2023, 14.5 percent of sales, with online at 17.6 percent. For bulky goods the cost of shipping a return is high. AR reduces the returns caused by the product not fitting the space and does nothing for colour disappointment or late delivery, so model it on that reason code and not on total returns.

Conversion figures exist and come with caveats. Shopify reported in 2020 that interactions with products carrying 3D or AR content converted 94 percent more often than the same kind of products without; that's merchant data and wasn't a split test. The Oakywood figures in Shopify's 3D e-commerce article, average order value up 458 percent in Germany, 232 percent in Switzerland and 84 percent in the UK, come from a configurator that added a made-to-order product line, so they describe new products more than a lift on the existing range.

Phasing the catalogue

Phase one is a pilot of 20 to 50 geometries: your top sellers by revenue, weighted towards large and hard-to-judge items, with full front-end integration and analytics. This phase exists to produce your own conversion number.

Phase two is the revenue core, 100 to 200 geometries, everything that contributes materially to revenue. By then the template work is done, so the marginal cost is modelling only.

Phase three is the long tail and the variants: material swaps across the existing geometries, then the slow-moving products, then configurators if you sell configurable ranges.

Spreading it this way keeps the first-year number small and lets the second year be justified from your own data.

Phase	Scope	What it buys you
1	20 to 50 top sellers	Your own conversion data, template built
2	100 to 200 revenue core	Most of the commercial benefit
3	Variants, long tail, configurator	Coverage and marketplace-proof pages

What to cut first

When the number has to come down, cut the long tail before quality. A model that looks wrong undermines the page it sits on.

Small products go next. A good photograph answers the question for anything the shopper doesn't need to fit into a space.

The configurator can move to a later phase, unless configuration is how you sell, in which case cut something else.

Cut variants before geometries: a product live in one fabric can take the rest of its fabrics later as material swaps.

Keep the analytics and the review time. Analytics is the cheapest line and the one that justifies next year's budget, and models nobody reviewed ship wrong and get pulled.

The business case

Modelling the products that generate most of the revenue creates an asset that answers the fit and quality questions blocking a large purchase and replaces part of the photography budget, and it can be measured against your own baseline within a quarter.

Before a number goes in the spreadsheet, send us one product and we model it at no charge so you can put it on a live page and measure it; pricing is on prices and scope on solutions.