What machine shops should know about manufacturing marketplaces

How manufacturing marketplaces make money

The business model is straightforward. A buyer uploads a part. The platform generates an instant quote using algorithms that estimate material, cycle time, and complexity. The buyer pays the platform. The platform finds a shop to make the part at a lower price than what the buyer paid, and keeps the difference.

That difference — the take rate — is typically 20–30% of the job value. On a $10,000 job, the platform might pay the shop $7,000–8,000 and keep $2,000–3,000. The buyer doesn't know what the shop was paid. The shop doesn't know what the buyer was charged. The platform sits in the middle of every transaction and controls the information on both sides.

This isn't inherently wrong. The platform invested in building the quoting engine, the buyer acquisition, and the matching infrastructure. They deserve compensation for that. The question is whether the value a shop receives — access to buyer demand — is worth what the shop gives up.

What shops give up

Margin

The most obvious cost. A 25% take rate on every job means a shop running $500K/year through a marketplace is paying $125K/year for lead generation. That's the salary of a skilled machinist, or the annual payment on a new CNC lathe. It's a real cost, and it compounds — the more work you route through the marketplace, the more you pay.

The take rate also creates pricing pressure. Because the platform is optimizing for the lowest cost supplier who can meet the spec, shops are implicitly competing on price against every other shop on the platform. The work tends to migrate toward the cheapest shop willing to take it. That's a race to the bottom, and it's very hard to reverse once you're in it.

The customer relationship

On most manufacturing marketplaces, the buyer is the platform's customer — not the shop's. The shop doesn't know the buyer's name until the job is awarded, and in some cases doesn't know it at all. If the buyer reorders, the platform can route that job to a different shop. The shop has no ability to build a direct relationship, negotiate terms, or develop the account over time.

For a shop that's spent decades building customer relationships through quality, reliability, and personal service, this is a fundamental change in how the business works. The shop becomes interchangeable — defined by price and capacity, not by reputation and expertise.

Pricing control

The platform sets the price the buyer pays. The shop can accept or decline the job at the price offered, but can't negotiate, can't explain why a particular feature costs more, can't offer alternatives. The nuanced conversation that happens between a shop and a buyer — "this tolerance is going to cost you, but if we change the approach here, we can save 20%" — doesn't exist. The algorithm doesn't have that conversation.

Quality reputation

If a shop does excellent work through a marketplace, the buyer attributes the quality to the platform, not the shop. The shop doesn't get the credit. If a different shop on the platform delivers poor quality on a subsequent order, the buyer blames the platform — but the original shop's reputation doesn't benefit either way. The shop's quality is invisible to the end customer.

When marketplaces make sense for shops

This isn't all negative. There are situations where marketplace work is a rational choice.

Filling idle capacity. If a machine is sitting empty and a marketplace job covers variable costs and contributes something to overhead, taking the work at a lower margin is better than running the machine dry. The key is treating marketplace work as fill, not as a growth strategy.

Breaking into a new process or material. A shop that just added wire EDM capability might take marketplace work at thin margins to build experience and samples before pursuing direct customers at full margin.

Geographic expansion. A shop in the Midwest that wants to test demand from West Coast buyers might use a marketplace to gauge interest before investing in direct sales and marketing in that region.

The problems start when marketplace work becomes the majority of a shop's revenue. At that point, the shop is dependent on the platform for its livelihood, has no direct customer relationships to fall back on, and is paying 20–30% of every dollar for the privilege.

The alternative model

Not every platform works this way. A network model — where shops are identified by name, set their own prices, own the customer relationship, and pay a flat membership rather than a per-job take rate — preserves the benefits of platform visibility without the costs of platform dependency.

In a network model, the buyer sees the shop's name, machines, capacity, and capabilities. The buyer contacts the shop directly. The shop quotes the work at their own price. The platform's role is matchmaking and visibility, not transaction control. The shop pays for being found, not for every dollar of work that results.

The Axhera Network works this way. Shops are listed by name with their machines and real-time capacity. Buyers find shops through process-specific content and a searchable directory. Introductions happen through the platform. Relationships happen directly. There's no take rate on any job.