Jewelry Configurator: Build Custom or Buy Off the Shelf?
Short answer: Buy if your product is simpler than you think it is. Build if your rules, your pricing or your production handoff are the reason customers choose you. Most jewelry businesses guess wrong in the same direction, they buy first, hit the ceiling in month eight, and pay for the build anyway on top of a year of subscription.
This post is the test we actually run with clients before recommending either. It takes about an hour to answer honestly, and it is the cheapest hour in the whole project.
Why the question is harder in jewelry than elsewhere
Configurator platforms are a mature category. There are good ones. They were largely built for furniture, apparel, signage, packaging and industrial equipment, and in those categories they work well, because those products share a helpful property: the options are mostly independent.
A sofa in blue fabric with oak legs is a valid sofa. So is the same sofa in grey with walnut legs. The options compose freely, and where they do not, a short exclusion list handles it.
Jewelry does not compose freely. A head sized for a one carat round will not take a six millimeter emerald cut. A two millimeter band will not carry the stones the customer just selected. A size 4 half eternity holds fewer stones than a size 10, so the two are different products at different costs. Metal price moves daily. And the output has to be manufacturable by a specific bench with specific equipment.
The gap between "product with options" and "product with a rules engine" is where off the shelf platforms stop and jewelry starts. That gap is the whole question.
The five tests
Answer these before you look at a single vendor.
1. The exclusion test
Write down every combination a customer could select that you would refuse to make. Not the ones that look wrong, the ones your bench would send back.
If that list is under about twenty rules and each one is a simple pairwise exclusion, an off the shelf platform will handle it. If your list runs to hundreds, or if the rules are conditional (this head takes this shape only between these carat weights at this metal thickness), you are describing a rules engine, and configurator platforms generally give you a rules table rather than a rules engine. The difference shows up the first time you need arithmetic instead of a lookup.
2. The price test
Is your price a stored number, or is it a calculation?
If you can put a price against every combination and update it monthly, buy. If your price is metal weight times today's rate, plus stones at their own rates, plus labor by complexity, plus finishing, plus margin that varies by channel and customer tier, then you need a pricing service, and the question becomes whether the platform you are considering will call yours. Some will. Many will only accept a price list.
This test alone decides a surprising number of cases, because a configurator that prices wrongly loses money on every order until somebody notices, and nobody notices quickly.
3. The handoff test
What has to happen when the customer clicks buy?
If the answer is "an order lands in our system and someone picks the right SKU", buy. If the answer is "a manufacturable file, a stone specification, a sizing instruction and a routing decision have to reach the factory as structured data", you are building, or at least building the half that faces production.
This is the test people skip, and it is the one that determines whether the configurator saves your operations team time or gives them a new inbox to retype from.
4. The differentiation test
Ask what a customer would notice if you had the same configurator as your three closest competitors.
If the answer is "nothing much, our product and our service are the difference", buy, and spend the saved money on the product. If the answer is "quite a lot, the configurator is how we sell", then it is a product, not a utility, and products get built.
Pairing a band against the engagement ring a customer already owns is an example of the second kind. Almost nothing on the market does it, it is the single most requested feature in that category, and it is not something you can configure your way to on a platform that has no concept of a second ring.
5. The catalog test
How many product families, and how fast do they change?
One family, stable, a few dozen options: buy. Six families, seasonal collections, options changing monthly: the ongoing cost of maintaining that in someone else's admin panel starts to exceed the cost of a system that takes your product data directly.
The scoring
Count how many of the five point toward building. Zero or one, buy without agonising. Two or three, the useful answer is usually neither pure option but a hybrid, and we come back to that below. Four or five, build, and be glad you asked before signing a two year contract.
What people get wrong in each direction
Buying when you should build. The failure is rarely dramatic. The configurator launches, it works, and then a set of small impossibilities accumulate. The rule you cannot express. The price that has to be recalculated manually. The order that arrives as a PDF. Each has a workaround, and eighteen months later the workarounds are somebody's full time job. The cost was never the subscription.
Building when you should buy. The failure here is faster and more visible. You commission a bespoke system for a catalog that a configured platform would have handled, you own the maintenance forever, and you discover that the thing you actually needed was better photography and a clearer size guide.
Assuming the platform's roadmap will rescue you. It might. It is not built for you, and jewelry is not its largest market. Never buy on a promise about the next release.
Treating it as one decision. It is at least four. The visual layer, the rules layer, the pricing layer and the production handoff can each be bought or built independently, which brings us to the answer most businesses in the middle actually need.
The hybrid, which is where most of them land
Buy the parts that are commodities. Build the parts that are yours.
In practice that usually means: a rendering layer that is bought or built on standard 3D libraries, because photoreal metal and stone is a solved problem and not a place to be original; a rules engine and pricing service that you own, running outside whatever the storefront is, because those are your business logic and they will outlive at least two storefronts; and a production handoff built once against your factory, because nobody else can.
The advantage of drawing the line there is that it is reversible. When the storefront changes, or the platform is acquired, or you move from a hosted store to a headless one, the layers you own move with you. The layers you rented were always going to be replaced anyway.
For more on how those four layers fit together, see our guide to what a jewelry ring configurator actually is, and on what each of them costs to build, see what a configurator actually costs.
The question to open with
Not "build or buy". Ask instead: which parts of this are our competitive advantage, and which parts are plumbing?
Nobody wins on plumbing. Buy that. Everything that is genuinely yours, own it, because the moment it lives in someone else's system it stops being yours.
What we do
At AMG Dynamics we run this test with clients before we quote anything, and we say buy when buying is right, which happens more often than you might expect from a company that builds these. When building is right we usually build the rules, pricing and production layers first, because those are the ones that outlive the storefront.
We work with jewelry brands, manufacturers and retailers from London, Delaware and Mumbai, across 3D configurators, visualization, workflow automation and AI systems.
Want a straight answer on which route fits your catalog? Send your product families, roughly how many compatibility rules you hold, and how your pricing is calculated to admin@amgdynamics.com.
If you decide to build, the shape of what we build is on our engagement ring configurator page.
Frequently asked questions
Should I build or buy a jewelry configurator? Buy if your options compose freely, your prices are stored numbers, and the order can be fulfilled by picking a SKU. Build if your compatibility rules need conditional logic, your price is calculated live from metal weight and stone selection, or the output has to reach your factory as structured production data. Most jewelry businesses land in the middle and are best served by buying the rendering layer while owning the rules, pricing and production handoff.
What is the main limitation of off the shelf configurators for jewelry? They generally provide a rules table rather than a rules engine. That handles simple pairwise exclusions well, and handles conditional geometry and size dependent stone counts poorly. Jewelry is unusual in that its options do not compose freely, which is the assumption most platforms are built on.
Can an off the shelf configurator handle live metal pricing? Some can call an external pricing service, and those are the ones to shortlist if your price is calculated rather than stored. Many accept only a price list, which means a manual update whenever metal rates move and a quiet margin loss in between.
Is a hybrid approach realistic? It is the most common good outcome. Rendering is a commodity and worth buying. Compatibility rules, pricing logic and the production handoff are business logic, and building them as services outside the storefront means they survive a replatform rather than being rebuilt with it.
How long before an off the shelf configurator hits its ceiling? When it happens, it is usually somewhere between six and eighteen months, and it shows up as accumulating manual workarounds rather than an outright failure. The warning sign is a person whose job has quietly become translating configurator output into something production can use.