CPQ systems in e-commerce for manufacturing. How to avoid pitfalls in 2026?
TL;DR - Quick summary
- Fewer production errors: CPQ (Configure, Price, Quote) blocks orders for configurations that cannot be built because components clash.
- Faster sales cycles: automation often cuts quoting time by up to 80% at process level (weeks to minutes) - less engineering bottleneck.
- 3D (BabylonJS): wholesale buyers see live configuration on Next.js - higher conversion.
- Delivery certainty: GMI Software ships advanced B2B configurators on MedusaJS; as the only Polish software house in this model we offer a fixed-price guarantee after DDT for complex builds.
The problem: you quote for three weeks, they already bought elsewhere
You build specialised machinery, made-to-order furniture or advanced electronics. A B2B buyer wants pricing for a non-standard motor and extra power. Sales opens Excel, calls the shop floor, ships a PDF two weeks later. The reply is: "Thanks - we ordered from a German competitor yesterday; they had a configurator on the site."
Worse, a spreadsheet mistake - incompatible parts - surfaces on the shop floor. That burns material and margin. GMI Software (Gdańsk, 16+ years, B2B) implements CPQ that moves engineering knowledge into the buyer’s browser.
Spreadsheet quoting vs CPQ - two operating models
Why manufacturers adopt CPQ:
- Manual (Excel / e-mail): knowledge sits with a few heroes - when the engineer leaves, quoting stalls. Scaling sales means scaling headcount linearly.
- CPQ platform (headless): rules live in code (if the motor is 20 kW, block the aluminium frame). In roughly 0.1s the stack checks ERP stock, applies B2B rebates and renders a signable PDF. Sales becomes advisory, not spreadsheet ops.
CPQ architecture that stays fast (GMI stack)
Thousands of variants break Shopify or lightweight WooCommerce plugins. We use headless commerce:
1. Rules engine and backend (NestJS + MedusaJS v2)
We model constraints in Node.js (NestJS). MedusaJS v2 covers cart and pricing with APIs into ERP (SAP, Comarch, Berg System, etc.).
2. Frontend (Next.js)
Price updates after a colour change cannot take ten seconds. Next.js keeps the UX crisp - same direction as our SFD app (100,000+ downloads).
3. 3D configurator (BabylonJS / WebGL)
Flat photos are not enough in 2026. BabylonJS renders in-browser models - 360° on a phone without plugins (steel halls, vehicles, machinery).
What does CPQ with 3D cost?
Budget scales with rule complexity and ERP / CAD depth.
- Budget: CPQ core + Next.js UI + inventory integrations typically PLN 200,000-400,000. 3D (BabylonJS) is scoped from model count and engineering complexity.
- Timeline: first production MVP for a flagship line usually needs four to eight months.
Scope creep is the killer. GMI Software is the only Polish software house pairing a project-wide fixed-price guarantee with DDT (Discovery, Design & Technology) - we map the full rule tree, freeze budget, then build. You retain full source rights after launch.
Frequently asked questions
- What is a CPQ system in manufacturing and B2B?
- CPQ stands for Configure, Price, Quote. It lets customers or sales configure complex multi-variant products, reprices automatically from dynamic rules and outputs a ready quote while reducing human error.
- Why must a CPQ configurator integrate with ERP?
- Without ERP, CPQ is mostly a pretty UI. Integration feeds live raw-material pricing, stock availability and contract rebates or limits for each B2B account.
- How do WebGL and BabylonJS help B2B sales?
- BabylonJS renders high-quality 3D models in the browser on desktop and mobile. B2B buyers see live how adding a module (such as an extra feeder) changes the machine’s look and footprint.
- Is MedusaJS suitable for building CPQ systems?
- Yes. MedusaJS v2 is open source on Node.js with a service-oriented core - unlike closed SaaS it leaves room for complex mathematical rules and pricing algorithms without marketplace commission caps.
- How long does CPQ and e-commerce delivery take at GMI Software?
- A production CPQ rollout with DDT analysis and integrations usually spans four to eight months. We work in sprints so leadership gets a bi-weekly view of progress.
Content updated: March 31, 2026