XO3D Studio · Buyer guide
Is a configurator
worth building?
This guide helps you decide whether to commission a configurator, and which kind. It is not a build tutorial. Configurator projects rarely fail on the 3D. They fail on arithmetic nobody did, a real-time budget nobody set, or a commerce handoff nobody owned. Six gates decide whether the idea is feasible, and all six can be answered before anyone models anything. This also covers the cases where the honest answer is that you do not need one.
The studio behind this guide
Where we sit on configurator work
A configurator platform renders what it is given, so somebody still has to build and finish the 3D assets it displays. XO3D builds the real-time product asset a configurator runs on and can build and embed the experience itself; a commissioned configurator can be handed over for you to host on your website, with hosting, updates and maintenance agreed in the brief. Come to us at the feasibility stage rather than after the arithmetic has been done, because the variant count and the real-time budget change what is worth building.
How XO3D compares to the other five supplier modelsStart with the route
Three routes to show options.
“Configurator” covers very different builds. Before the six gates, decide which of three routes fits. A stills library shows a bounded set of combinations. Layered images swap prepared renders on a fixed view; the Acumen configurator in the questions below works this way, and Rollink built its own colour selector from XO3D imagery. Real-time 3D draws the product live in the browser. These are scoping categories, not a promise that one route is always cheaper or faster.
| Route | Suits | What you need | When options change | Who owns the interface |
|---|---|---|---|---|
| Stills library | A bounded set of combinations shown from agreed angles | The list of combinations to show, the angles and the source for each finish | Each new option needs its own renders for every angle | Your site or content team places the images |
| Layered images | Predefined visual options that swap on a fixed view | Rendered layers for each option, plus the rules for which layers may combine | New layers are rendered and the rules are updated | Whoever builds the selector interface, with the product owner approving the rules |
| Real-time 3D | An experience where free rotation or live controls answer a real buyer question | A web-ready model, materials for each option, the option rules and a device budget | New materials or modelled parts are added and the experience is retested | The developer or platform that builds and hosts it |
Compare interactive formats for viewers, configurators and AR side by side.
Answer these first
The six gates.
Answer these questions before development. Each can change the appropriate asset, interface or delivery approach.
The variant arithmetic
Multiply your options together before anything else, but treat the answer as a ceiling. Six colours, four fabrics and three sizes give 6 × 4 × 3 = 72 combinations only if every option can be combined with every other. Real ranges have rules, and the valid set is usually smaller. Each camera angle or output you need is a further decision on top. The useful questions are which combinations customers actually need to see and which ones the business will actually sell.
What varies, and whether anyone cares
Colour and material swaps are usually lighter work than geometry changes. A different handle, an extra module or a longer chassis each need their own modelled variant, which has to be scoped separately. It is worth checking which options customers actually choose before paying to make all of them interactive. A configurator is sometimes specified around the full catalogue when a handful of combinations carry most of the demand.
The real-time budget
A hero render can carry far more detail than a real-time asset, because it renders once, offline. An interactive asset has to load and run on the devices your customers use. That is a different asset built to a different budget, not the film model made smaller. Geometry, textures, requests and total download all need a limit, set for this project with the technical owner and tested on real devices.
Where it runs, and what your site allows
The experience has to sit inside a real page on a real platform. That means page weight budgets, a content security policy that will permit whatever loads it, mobile layout, and whether your commerce platform can carry the chosen configuration through to the basket and the order. That last one is a commerce question rather than a 3D question, and it is the one most often discovered late.
Who builds, hosts and maintains it
Someone has to build the experience, put it live, and still own it in eighteen months when a browser update changes something. That can be XO3D, your own developers or your platform vendor, and it should be settled before production rather than after. Maintenance is the part that gets left out of the plan.
What the asset does afterwards
The master build behind a configurator can also feed stills, film and AR, if it is planned for that. Whether it can depends on the geometry, the materials and where each output is going. Agree at the outset whether the model is built for reuse or only for the viewer, and confirm what you own at the end so extending it later is not a fresh negotiation.
The arithmetic, with rules applied
A fictional sofa comes in six colours, four fabrics and three sizes: 72 combinations if everything combines. Now suppose two of the fabrics are not made in the smallest size. That removes 6 × 2 × 1 = 12 combinations, leaving 60 that can actually be ordered. Each view you show is a further multiplier, and each output, such as a listing image or a selector layer, is its own scoping decision. Seventy-two possible choices is not seventy-two necessary renders, a fixed cost or a point at which one route becomes cheaper. The product owner provides and approves the real compatibility rules.
Where it runs: the selection and the order
A visual selection is not an order. The picture shows a combination; a commerce or quote system records it. Ask who maintains the product IDs, the valid combinations and the prices, and whether the accepted selection passes to a basket, a quote or neither. Then ask where that selection is recorded and who checks it before anything is made. A screenshot of a configured product is not an approved production specification. Your commerce system and its implementation owner stay responsible for the commercial and manufacturing data; agree ownership of the handoff before the build starts.
A device test plan instead of a rule of thumb
Name the representative devices and browsers your customers use, the connection conditions to test, how the experience should load and what a visitor sees if it cannot run. Set a budget for geometry, textures, requests and total download with the technical owner, then test the actual experience and record the results. Whether every option loads at once depends on how the experience is built. Consider web and AR file-size constraints when you set the budget.
On what goes wrong when a configurator has too many options
The main issue is file size. Every extra model and every extra colour has to load, and it gets heavy on the network very quickly. Phones can't load it fast. It all comes down to file size and loading.
Honest disqualification
When you do
not need one.
A configurator needs a maintenance owner and a process for updating product rules, assets and supported devices. Four situations where something simpler is the better answer.
- A 360 viewer would do
- If customers mainly want to see the product from every side, a turntable answers that and is far simpler than a configurator to build, run and maintain.
- A stills set covering the real demand
- Where a handful of combinations carry most of the orders, rendering those properly and showing them well may answer the buyer’s question without an interactive.
- Nobody owns the integration
- If there is no development capacity on your side and no agreed owner for the commerce handoff, a configurator will stall at exactly that point. Fix the ownership question first.
- The range is about to change
- If the range is being revised, note which inputs are stable and which decisions can wait. Building interactive variants for parts that are about to change can mean building them twice.
A focused stills set or a 360 viewer is enough when it answers the question the buyer is actually asking: what the product looks like in the finish they want, or what it looks like from every side. Whether it is the better choice for your range depends on a scoped comparison of the outputs you need, not on a rule. If the range or its rules are changing, write down which inputs are stable now and which decisions you are deferring, so the simpler route can grow later without starting again. This is a planning recommendation, not a rule to wait in every case.
If one of those describes your situation, when do you need interactive 3D compares the lighter formats and what each actually supports.
Write it down
A feasibility record.
Use one row per gate to record what you know, who owns it and what is still open. The entries below are prompts, not answers. You can add a pass, investigate or stop label to each row as a discussion aid; it is not an automated decision or an engineering certification, and the decision stays with you. A stills set or a viewer stays on the table throughout.
| Gate | Evidence available | Owner | Open decision | Next action |
|---|---|---|---|---|
| Product and rules | Option list; which combinations are valid | Product owner | Which combinations customers need to see | Approve the valid set |
| Visual route | The buyer question the experience answers | Marketing lead | Stills, layered images or real-time 3D | Choose a route, or a simpler format |
| Source assets | CAD, finish references, existing renders | Product owner | Which parts are final and which may change | Mark stable and provisional parts |
| Destination and devices | The page, platform and representative devices | Technical owner | Budget for download and loading | Agree a device test plan |
| Integration | Where the selection goes after the choice | Commerce or implementation owner | Inform only, basket or quote | Name who validates the selection |
| Maintenance and handover | Who updates options, rules and the build | Named owner on your side | What happens when the range changes | Agree the handover and update route |
Before you brief
What to bring
to the first
conversation.
- The option list: every colour, material, size and component variant, and which of them change geometry rather than just finish.
- What sells: which combinations actually carry your orders. This usually changes the plan.
- The page: where it will run, on what platform, and any page weight or content security policy limits your site enforces.
- The commerce path: whether the chosen configuration needs to reach the basket, a quote, or just inform the decision.
- The owner: who implements it, and who still owns it in eighteen months.
Who does what across those last two is set out on integrations.
Related
Where to
go next.
Common questions
Straight
answers.
Do I need to build a 3D configurator?
Not always. Start with the question your buyer needs answered. A stills set can cover a bounded range of combinations, layered images can swap predefined options on a fixed view, and real-time 3D suits an experience where free rotation or live controls matter. Pick the simplest route that answers the question.
How many product variants before a configurator is worth building?
There is no single number. It depends on which combinations are actually valid, how many views each channel needs, and whether the variation is material or geometric. Multiplying the options gives a ceiling, not the set you need. List the combinations customers actually choose and the outputs each one needs, then compare a scoped stills set, layered images and real-time 3D against that list.
Can you use the same 3D model for the launch film and the configurator?
The same underlying build can often serve both, if it is planned that way from the start. The same delivered asset, no. A film model renders offline and can carry far more detail; an interactive asset has to load and run on the devices your customers use, so it is derived from the master build to a real-time budget rather than being the film model made smaller.
Who builds and hosts the configurator?
It varies by project and it should be agreed before production. For Acumen, XO3D built an image-layer furniture configurator, which swaps prepared renders rather than drawing the product in real time, and its web developer, Lemuel, handled integration into the client website. Later option updates were not included. For Rollink, the client team built its own colour selector on its website from XO3D imagery. Agree who builds the interface, integrates it, hosts it and handles later updates; an asset commission alone does not settle those responsibilities.
Will it work on mobile?
Plan for it rather than test it at the end. Name the representative devices and browsers, the connection conditions, how the experience loads and what visitors see if it cannot run. Set a budget for geometry, textures, requests and total download with the technical owner, then test the actual experience on those devices and record the results.
What happens when our product range changes?
Material and colour options are usually straightforward to extend, because they reuse the existing build. New geometry is a new modelled variant and needs scoping as such. If a range revision is already planned, it is often better to build the stable part of the range first rather than pay to make variants interactive twice.
Start with the arithmetic
Send us your
option list.
Tell us the variants, where it needs to run and who owns the front end. We will tell you whether a configurator is the right answer, and say so if it is not.
Talk through your rangeYour next project
Test the decision before the interface.
A configurator should help a buyer make a meaningful choice. Define valid combinations, the product source and a test device first. The G-Lab film below demonstrates product detail, not a delivered configurator.
A route to consider
Product interactives
A product experience scoped around a useful interaction, supported devices, model performance and a clear fallback.
Bring these three things.
- Which choices can the buyer make?
- Which combinations are invalid?
- What devices and fallback must be supported?
Start with an outline. We review the product, intended outputs and timing before agreeing the production scope.