ODM manufacturing: owning the design and selling it under other people's brands

Engineering becomes a fixed cost you must keep fed

Holding the design means holding a development team whose salaries continue whether or not a brand signs. Recovery comes through unit margin spread across every customer that adopts the platform, which makes the number of adopters matter more than any single order size. The commitment is to a roadmap: a base product, a refresh cadence and a support horizon for units already in the field. Committing a plant to that rhythm means engineering capacity, not machine capacity, is usually the first constraint the business meets. Pausing development to save money is rarely reversible either, because the specialists leave and the platform ages while nothing replaces it.

One platform, several badges

Design work concentrates on a base architecture with defined option points — housing colour, connector variant, firmware branding, packaging. Demand suits the model when several brands want broadly the same function with cosmetic separation and none of them wants to fund development alone. Stock follows the same logic: common sub-assemblies are held deep and finished variants shallow, with differentiation delayed until an order names the customer. Minimum quantities per variant then become the negotiating point, because each badge adds tooling, artwork and a slow-moving stock line that somebody has to carry.

Approvals in your name are an asset and a leash

Safety files, electromagnetic compatibility reports, type approvals and test evidence sit with the design owner. That is a genuine asset: a brand adopting the platform inherits a route to market without repeating the testing, and the certificates raise the price a newcomer must pay to copy you. It is also a leash. Any change to a critical component or supplier can re-open testing, laboratory slots are scarce, and the cost lands on the factory rather than the customer. Access to instrumented test capability, in-house or contracted, belongs in the capital picture alongside moulds and assembly cells.

The bill of materials and its problems belong to you

Choosing every component means owning obsolescence notices, allocation shortages, alternate qualification and cost-reduction work across the whole range. Purchasing leverage improves because volumes from several brands aggregate onto the same part numbers, which is one of the model's real advantages over building inside a customer's approved vendor list. The corresponding exposure is that a single end-of-life notice can strike every customer at once, so the design team needs a second source qualified for anything with a short market life, and the commercial team needs language permitting a technical change without renegotiating each account separately.

How the design gets taken away from you

The characteristic failure is blunt: a customer buys enough volume to justify taking the design elsewhere, or a competitor reverse-engineers the platform once the certificates make the route obvious. Registered designs, patents where they genuinely apply, firmware that resists lifting and a refresh cycle faster than a copyist can follow are the practical defences. Growth stalls when the engineering backlog fills, when variants proliferate past what the item master can carry sensibly, or when two customers with overlapping channels both demand exclusivity. Lifecycle management, variant configuration and disciplined revision control stop being optional at that point.

Frequently asked questions

Should we let a brand customer put its own name on our certification?

Transferring or extending a certificate makes the brand the responsible party in the eyes of market surveillance authorities, which is often exactly what they want and rarely what serves you. Keeping approvals in the factory's name preserves the ability to sell the platform to others and to make technical changes without collecting signatures. A licence letter permitting the customer to reference your test evidence usually satisfies their compliance team without giving away the underlying asset.

How many customers does an original design platform need before it pays for itself?

Think in terms of development recovery per unit rather than a customer count. Total design, tooling and approval spend divided by the volume the platform can realistically win over its market life gives the amortisation each unit must carry; if that figure eats a large share of the target price, the platform is too narrow. Platforms recovering their cost from a single account are effectively bespoke development and should be priced and contracted that way.

What stops a brand customer from asking for design changes we cannot afford?

A published boundary between the platform and paid engineering. Say plainly what is configurable at no charge — colour, label, packaging, a firmware string — and treat anything touching the certified design as a chargeable project with its own quotation, lead time and re-approval budget. Without that boundary every account drifts toward bespoke development, engineering capacity disappears into customer-specific work, and the economics that justified owning the design quietly evaporate.