Product configurators: encoding what you will build, not everything you could

The rule set is engineering knowledge written down

In most engineered-to-order businesses the knowledge of what fits with what lives in a few experienced heads and a spreadsheet of exceptions. A configurator forces that into explicit rules, and the forcing is most of the value: the exercise surfaces contradictions, forgotten restrictions and options nobody has actually built for years. It also exposes how much of the current answer depends on one person's judgement. Expect the first pass to be uncomfortable, because writing down a rule invites someone to challenge it, and several long-standing restrictions will turn out to have no remaining basis.

Impossible, expensive and unsupported are different constraints

Rule sets conflate three distinct categories. Some combinations are physically impossible and must be blocked outright. Some are possible but carry cost, lead time or engineering effort that the price must reflect, and blocking them loses business you would have wanted. Some work fine but are not supported by service, documentation or spares, which is a commercial policy rather than a technical fact. Modelling all three as hard blocks makes the configurator an obstacle that sales route around. Modelling them separately lets a valid but costly configuration proceed with the right price and the right internal warning.

From a configured order to a bill and a routing

Generating a specification is useful; generating the production structure is where the plant benefits. That means rules producing the bill lines, quantities, operation sequence and any option-driven work content, so an order arrives with something manufacturable attached rather than a description an engineer must interpret. The design decision is how much becomes real data. Creating a stored part number for every possible combination is unmanageable in a wide configuration space, while creating nothing makes traceability and repeat orders awkward. Most businesses land on identifying the configuration itself and generating structure per order.

Selling configurations the plant cannot make

The frequent failure is a sales-facing configurator built for quoting speed and disconnected from the manufacturing rules. It produces attractive quotations for combinations that engineering then declines, which converts a sales win into an internal argument and a customer disappointment. Where the two must be separate systems, the manufacturing rule set has to be the source and the sales tool a derived copy on a controlled refresh. Any locally added sales rule is a future escape. Track how often orders need engineering intervention after configuration, since that rate tells you how far the two have drifted.

Rule maintenance is a permanent job, not a project task

Every engineering change, discontinued component, new supplier and revised specification potentially touches rules, and nothing in the system announces which ones. Without a named owner and a link from the change process into rule review, the set decays: blocked combinations that are now fine, permitted ones that no longer are, prices attached to options that changed. Decay shows up as growing manual override use, which is the metric worth watching. Once the shop floor treats configurator output as a suggestion, the investment has effectively been written off while still being paid for.

Frequently asked questions

Is a configurator worth it if we only sell a handful of variants?

Probably not as a system. A small, stable variant set is better handled as a set of defined products with their own structures, which is simpler to plan, cost and quote. The case for a configurator strengthens when the combination space is too large to enumerate, when quoting depends on scarce engineering time, or when errors in specification are reaching the plant regularly. Combination count matters less than how often a person is currently making a judgement call.

Should every configuration get its own part number?

Only where you will genuinely need to refer back to it as a product: repeat orders, spares support, regulatory registration or long service life. In a wide configuration space, creating a number per combination fills the item master with entries used once and never again, which slows every search and every report. A configuration identity recorded against the order, with the generated structure retained, usually gives the traceability without the permanent data burden.

Should sales or engineering own the configurator rules?

Engineering owns what is technically valid, commercial owns what the business is willing to sell and at what price, and the system needs both layers with different approvers. Giving sales control over technical constraints produces orders the plant cannot build. Giving engineering control over commercial availability produces a tool that blocks profitable work for reasons nobody can explain to a customer. Separate the layers and the ownership argument mostly disappears.