E-commerce & Product Models

Made-to-order manufacturing business

You build each item only after it is ordered, to the buyer's own drawing, billing per job at a price and delivery date you quote up front: a machined bracket, a custom cabinet, a one-off sign.

  • Advanced
  • $5K–$25K
  • Low risk
  • 6+ months to first customer

These bands place this model against the other 171 in the catalog so comparing them works — orientation, not a quote for your situation or your area. Figures that carry a source are on the Examples tab.

Why this stability rating: Demand is genuine and repeats, since every product company needs parts, but nothing is contracted: revenue is job-by-job and tracks customers' R&D and capital budgets. Protolabs' revenue was roughly flat in 2024 and its customer count fell in 2025 even as revenue grew.

  • Asset-heavy
  • Local
  • Part-time friendly
  • Sales-driven
  • Inventory

Often fits: People who like tangible products and marketing, can hold discipline on numbers, and have (or can save) modest capital they can afford to park in inventory.

Often doesn't fit: People with no cash cushion, allergic to details like shipping tables and return policies, or hoping ads are a money printer.

The simple explanation

Someone wants a thing; you sell them the thing. E-commerce is the most legible model in business, but the simplicity is deceptive. Winning depends on margins after every hidden cost (shipping, returns, fees, ads), on conversion, and on whether customers come back. The product is the start. The machine around it (offer, funnel, fulfillment, repeat purchase) is the business.

A simple hypothetical example

Illustrative — invented to show the shape of the E-commerce & Products pattern. No real company is named, and no figure in it is data. The real, sourced companies for this model are on the Examples tab.

You notice dog owners improvising seat covers that don't fit. You source a better-designed one, brand it well, and sell at a healthy markup over landed cost. Ads bring the first customers; reviews and repeat accessories bring the profit. The winner here isn't the cover. It's the math: acquisition cost comfortably below first-order margin, and a customer who buys twice.

A closer look at made-to-order manufacturing business

Made-to-order inverts the inventory problem: nothing exists until it is sold, so there is no dead stock, no forecast and no clearance sale. There is also no volume runway, because every job starts from a fresh setup. All the margin lives in the quote, which is why the winners automated that step rather than buying more machines: Protolabs' 10-K describes automated quoting wired straight into fulfillment, protected by 60 patents, and it earns a 44.5% gross margin on $533 million of revenue. That is roughly ten points above the 34.7% Xometry earns brokering comparable work, because Xometry pays an outside shop to actually make the part.

This is the fork in the model. Own the machines and keep the margin but carry the fixed cost and the idle hours; broker the work and stay asset-light but live on a spread. Xometry grew marketplace revenue 30% to $630 million in 2025 and still lost $61.7 million, and MISUMI valued Fictiv, the same brokered shape, at $350 million. The growth mechanic nobody expects is that you win by serving fewer customers, not more: Protolabs' customer contacts fell from 51,552 to 48,415 in 2025 while revenue rose, because revenue per contact climbed 13.3% to $11,012. Prototype buyers are cheap to win and worth little; the same engineer coming back with a production run is the whole business.

What breaks first is estimating. Price a part before you have ever made it and every error is absorbed by you rather than passed on. The classic failure is a shop that is busy, well-liked, and quietly losing money on the third of jobs it under-quoted, because scrap, rework and setup time were guessed rather than timed. And with 18,734 U.S. machine shops competing, nobody wins on capability: they win on quote speed, on hitting the promised date, and on being findable when an engineer with a CAD file needs a part in nine days.

How money moves through this model

Who pays: Consumers (or businesses) buying online

What they pay for: A product that solves a problem or scratches a want, plus the trust to buy it sight unseen

What creates profit: Price minus landed cost, fees, shipping, returns, and the ads it took to win the order

  • Customer
  • Offer
  • Made-to-order
  • Costs
  • Profit

What makes this model hard

The honest difficulty: everything costs a little more than the spreadsheet said. Ads underperform, returns bite, platforms take their cut, and inventory ties up cash you can't spend twice. The sellers who survive are the ones who know their unit economics cold before scaling spend.