Design It Yourself, Outsource It, or Hire an Engineer: How to Decide

Most articles on outsourcing mechanical engineering vs keeping it in-house treat it as a two-way choice. For a small company or a person with a product idea there are four options: design it yourself, outsource it, hire an engineer, or train someone you already employ. Each one is right some of the time.

I run a contract engineering firm, so I have an obvious bias. I will also tell you where a contract engineer is the wrong call, because a client who should have hired in-house is not a client I keep.

The short answer

When to design it yourself

Do it yourself when being wrong costs you almost nothing. A bracket you can 3D print tonight and reprint tomorrow does not need an engineer. Neither does a first mockup whose only job is to show you whether the idea is worth pursuing.

Early on, the questions are yours to answer anyway. Nobody else knows what the product should do, who it is for, or what it has to cost. Cardboard, a hand sketch, and a printed part will answer those faster and cheaper than I can.

Stop designing it yourself when any of these becomes true:

• Two or more parts have to fit together and the fit matters.
• You are about to pay a shop to cut metal or build tooling.
• A machinist is asking questions about your drawing that you cannot answer.
• The part gets coated, plated, or heat treated and you have not checked what that does to the dimensions.

Past that line, mistakes stop costing an evening and start costing a production run. The reasons are covered in what a tolerance stack-up is and CNC machining vs 3D printing for prototypes.

When to outsource mechanical engineering

Outsourcing fits work that comes in bursts. A new product, a redesign, a drawing package for a quote, a problem that showed up on the shop floor. You need forty hours this month and none next month. A salaried engineer is a poor match for that pattern, and a contract engineer is a good one.

It also fits a specific skill gap. Plenty of shops have people who can model a part but nobody who can run a stack-up across an assembly, apply GD&T that a shop can inspect against, or review a design for manufacturability before it is released. Those are narrow, high-consequence tasks. Buying them by the hour is usually cheaper than hiring for them.

The work I get asked for most often falls into four groups:

• CAD design for parts and assemblies that are headed to a real manufacturer.
• Drawing packages and GD&T that a shop can quote without calling you five times.
• DFM review on a design someone else created.
• Prototyping, from printed parts through machined metal.

The trade-off is knowledge. A contractor learns your product and then leaves. You limit that by insisting on deliverables you own: native CAD files, a full drawing package, and written notes on why the design is the way it is.

When to hire an in-house engineer

Hire when the work is steady. If you would have real engineering tasks for someone most weeks of the year, a contractor becomes the expensive option and an employee becomes the cheap one.

Hire when the engineer needs to be physically present. Supporting production, walking the floor, catching problems at the machine, and answering the assembler's question in five minutes are jobs for someone who is there every day. A contractor answering email the next morning does not do that job.

Hire when the product is the company. If your design is your competitive advantage and it changes constantly, the knowledge of why it is built the way it is should live with someone on your payroll.

When to train the engineer you already have

This option gets skipped, and it is often the best one. Many small shops already employ someone who could do the work: a machinist who has been fixing bad drawings for years, a drafter, a technician who knows the product better than anyone.

That person already has the hardest thing to teach, which is how your product behaves in the real world. What they usually lack is formal and specific: GD&T, tolerance analysis, more depth in CAD. Those can be learned.

Training works when you give it three things:

• Protected time. Hours set aside every week that production cannot take back.
• Tools. A proper CAD seat and a course or two.
• Review. Someone experienced checking their work until they stop making the expensive mistakes.

The third one is where the options combine. A few hours a month of outside review on your own person's drawings costs far less than outsourcing the whole job, and your person gets better every time.

Training fails when it is expected to happen in the gaps between production work. There are no gaps. It also does not help if the deadline is five weeks out. Outsource that job and train for the next one.

What each option actually costs

Here are my numbers, since vague ranges do not help anyone decide.

Every project starts with a capped estimate: a fixed maximum cost to reach a defined milestone, billed hourly against that cap. You see the deliverables at the milestone and decide what comes next.

To compare that with an employee, use the full cost of the hire and not just the salary. Add benefits, payroll taxes, a CAD license, a workstation, recruiting, and the months before a new engineer is productive on your product. Then divide by the contract rate.

As an example, if an engineer costs your company $120,000 a year all in, that buys 960 contract hours at $125. That is about 18 hours a week, every week. Put your own number in, but the shape of the answer holds:

• If you need well under that many hours, outsourcing is cheaper and you carry no cost in the slow months.
• If you need more than that, year after year, hire.
• If you are near the line, train the person you have and buy outside review by the hour.

Designing it yourself looks free and is not. It costs your time, and if a mistake reaches the machine shop, it costs scrap and a second run.

Do you need a mechanical engineer for a product idea?

Not on day one. First find out whether anyone wants the thing. A rough mockup and ten conversations with possible buyers will tell you more than a finished CAD model.

Bring in an engineer when the idea has to become parts that somebody else will make. That is the point where material choice, tolerances, and manufacturing process start to set your cost per unit, and where changes get expensive quickly. A proof of concept at 4 to 20 hours is usually enough to learn whether the design is feasible and roughly what it will cost to build.

When a contract engineer is the wrong call

• You need someone on the floor daily. Hire.
• You have steady, full-time engineering work. Hire. You will pay less and keep the knowledge.
• You only need existing sketches redrawn in CAD. A drafter costs less than an engineer, and that is the right tool for that job.
• You cannot describe what done looks like. Without a defined milestone, hourly work drifts. Work that out first, even roughly.
• You have a capable person and no deadline. Train them.

If you are unsure which of these describes you, describe the project and I will tell you which option I would pick in your position, including when the answer is not me.