Editor’s note: Rui Ma is a partner with 500Startups based in Beijing.

I have to admit, before I decided to fly to Southern China this past week and dig deeper into the hardware ecosystem, I knew very little about anything to do with manufacturing. Sure, I looked at hardware-related investments, as every investor does these days, and tried to apply concepts of unit economics to the products, but I could not assess very intelligently whether the companies’ projections made sense, or were just a ton of acronym-filled BS.

I also struggled to know what kinds of challenging questions to ask. Especially at the seed stage, where companies are often coming to me fresh off of the success of a Kickstarter campaign and there isn’t a ton (OK, hardly any) of user data (or data of any kind) to paw through, I found myself mostly nodding in agreement with what the founders would say. Yes, that sounds like a reasonable assumption, I would think. But what did I really know?

I have no idea how long or how much it costs to do anything, or what happens when things go wrong, as they are apt to do. So I sat down with a few manufacturing experts who have either launched new consumer electronics products from scratch on their own, or have had substantial roles in them, and did nothing but ask questions.

This is the most important thing I learned, which goes against the grain of nearly every hardware business plan I’ve ever seen: It is advisable to budget 12-14 months for delivery of a new product, and here’s why.

Suppose you are building a product that may have some parts that are already available on the market, i.e. a battery with a solar panel charger, but is mostly a new concept, and not a simple commodity like a USB memory stick with flashlight. Then, here are the steps it might take for that product to be actually made at scale.

I am going to assume here that a fully working prototype designed for manufacturing (DFM, i.e. using manufacturer-friendly components and generally having specifications that are within reach of mass production) already exists. This, by the way, is not an assumption that investors should make, because not all makers or engineers know to optimize for that from the very beginning; in fact, that may not be very easy to do for new-fangled products.

Now, supposing your prototype is roughly DFM, what actually happens? Let’s peek inside the black box of a factory’s process.

1. You find factories (hopefully not just off of Alibaba, although it has served many simple customers well) and send out a Request for Quote (RFQ). Assuming you’ve been looking around all this time, and know roughly what you need, and/or have a few trusted introductions on the table, this may take you 1-2 weeks.

2. You do on-site visits to ascertain which factory (as a seed-stage startup, you’re probably working with only one) to choose. You sign a contract to work together. Budget another 1-2 weeks.

3. The factory begins the engineering drawings and whatever other work they need to do to get your product made. These are costs you have to pay upfront, by the way, and they are called non-recurring engineering, or NRE, because these are one-time costs incurred in the design, development and testing of a new product. This will probably take four weeks.

4. Now you’re ready to have the factory attempt to make the first batch of products, or the engineering validation test (EVT). This is typically a batch of 5-10 items — mostly hand-made and very rough pieces — that is all about manufacturability, form factor, and general product quality. Did the factory really understand your product specs? Does it look and feel like what you’ve done in your garage or lab or makerspace? Another 4-6 weeks will have passed doing this and performing some basic tests.

5. You’re happy with the EVT samples, and now it’s time to get started on tooling. Chances are you have special molds you need for your product, and at least for something like plastic, the factory will need to go order the steel, have a tooling company make the molds, and this money you will likely need to pay upfront as well, because it is often outsourced and besides, no one else will be able to use this. Hopefully you already have the steel ordered, and it’s just the two weeks for the tooling that you need at this juncture. If you were really aggressive and were sure you knew what you wanted (as in the case sometimes of the second generation of a product you already have been manufacturing successfully), you might do this concurrently with the engineering run. But for first-timers, doing this in parallel would be ill-advised. You’re out another 4-6 weeks.

6. And now for the design verification test samples, which is typically a 20-50 unit run that you are doing with the factory’s machinery in place as they try to figure out how to produce from the tooling that you have just had made. You should be rigorously testing the products that come off the line, ideally 1-2 months, but a minimum of two weeks would be advisable. The entire process may take an average of 4-6 weeks, depending on product.

7. Most products will need multiple of these small-batch runs until it is feasible to ramp up production. That’s because it is highly unlikely that the product performs exactly as you think it will, particularly as you stress-test it under different conditions. This is called the design refinement stage where minor, not major changes are being made to the product. You’re probably looking at multiple iterations, but let’s optimistically say that you only have to do one more trial because your product isn’t super complicated, and you’ve just been generally one damned lucky team. Move forward another 4-6 weeks. In parallel, if your product is relatively set, you might want to save some time by getting your various certifications, such as UL, CE, CSA, or perhaps ETL, depending on what you’re selling and where you are selling it into. You will also want to have “in-the-field” beta testers, i.e. people not intimately involved with the product to test it out, so that no obvious flaws are overlooked.

8. Finally, you’re at the pilot (or pre-production) run stage. This is where the factory has your product on the line, and will typically make one-tenth of your order quantity just to make sure that its own processes run as predicted. This is where you can relax, because it’s now the factory’s job to be efficient, to get the workflow just right and to get it done. However, for all intents and purposes, the products now coming off the line are customer-ready and deliverable. The remainder of the products will be produced using the learnings in efficiency and line management that the factory has learned from the pilot run. Tada! This may take the least amount of time, depending on quantity. Let’s say 1-2 weeks for a typical product.

So how much time has elapsed? If you have been doing the math all along, we are talking about at least six months for a relatively simple product that only goes through one real iteration. How likely is that? Tough to say, but probably not the majority of cases, which is why only 16 percent of top Kickstarter campaigns supposedly deliver on time.

The majority of teams, especially those who are stepping into the manufacturing center of the world for the first time, can probably only dream of such a smooth process. Add one or two more 4-6 week iterations, an additional month or two for testing and certifications and packaging, and we are easily at 12-14 months from prototype to delivery of an interesting but manufacturable new hardware product.

And that’s for folks who start right away and can get immediately plugged into the supply chain, and assuming no events such as worker strikes, component shortages, holiday production crunches, or factory bankruptcies (what, you think all the factories you work with are flush with cash?). Even in my relatively short two-year lifespan as an investor (as well as Kickstarter backer), I know that such seemingly generous deadlines slip … all of the time.

Let’s face it, hardware is a different animal, and both investors and entrepreneurs alike should strive to understand more about what the potential pitfalls are of working in this space. In this article, I’ve only covered the basic timeline of working directly with a factory.

I’ll leave you aspiring hardware entrepreneurs with one piece of advice: At the very minimum, have your CTO or chief product or engineering gal/guy on the ground wherever your factory is, at least for the first 2-4 weeks, and at minimum once a month after that. And it is crucial that they are on-the-ground for the pilot/production run to ensure that all the changes actually made it from product development onto the factory floor.

Most factories that you’re likely to be working with will let you have a desk in their facilities, and most will have a few decent English speakers, so don’t worry so much about communicating or having a place to stay. If you’re planning to manufacture in China, there is no excuse for not being here.