Would you switch to domestic PCB assembly if it actually was faster?

I run a tiny side business making replacement control boards for old CNC retrofits, so I've lived both sides of this. When I'm doing the first cut on a new board, I almost always go with the cheapest overseas option and just order five instead of one, because the per-board price barely changes at that quantity and the shipping wait doesn't bother me. I use that waiting time to review the layout again with fresh eyes, which honestly catches more mistakes than any inspection would. Where it falls apart is when I've got a customer waiting and my design has a stupid error, like the time I mirrored a connector footprint and didn't catch it until boards were in hand. Three week respin plus another three weeks shipping and I lost the client. After that I started paying extra for the faster local option on anything customer-facing, and I treat the premium as insurance rather than a manufacturing cost.

On the automated assembly question, I'm more worried about the opposite problem than a human missing something. Machines don't get bored, but nobody at an automated line is going to call you and say 'hey, this silkscreen looks wrong, did you mean it?' A human eyeballing a panel catches things that look off even when they pass the checks. I'd happily pay for a service where someone actually looks at my Gerbers before spinning the line, and I'd be suspicious of the cheapest fully hands-off option for that reason.

For my part, the turnaround question flipped for me once I stopped doing pure prototyping. I build a small line of weather station accessories, and the year I tried to save money with an overseas fab on my second revision, a batch came back with the component footprints mirrored on one channel. Two months gone, plus the cost of refunding three customers. Since then I pay more per board locally and the premium has paid for itself twice over, mostly in mistakes I didn't have to eat.

On the automation worry, my experiance runs the other way. My local shop runs optical inspection on every panel and they caught a hairline solder bridge on an order last spring that I would never have spotted by eye. Humans get tired at the end of a shift. Machines don't, though I still do my own visual check on arrival because nothing replaces eyeballing your own work. Where I'd push back on the original post is the framing that speed only matters during iteration. For a small seller, predictable timing is worth almost as much. When a board arrives when the supplier says it will, I can promise customers real ship dates instead of hedging with 'about four weeks, probly.' That predictability is honestly the thing I'd least want to give up, even more than the raw days saved.

I agree with the sentiment that you should view the local cost as risk mitigation rather than just a manufacturing price. It feels like buying peace of mind when your reputation is on the line. I ran into this issue myself with a small run of sensors. The automated visual check passed everything, but one component was sitting just slightly off the intended pad. It worked initially, but it was going to have horrible thermal performance later on. I only caught it because I was manually inspecting the board before it went into the unit. That slight shift would have been missed by a purely automated optical inspection system.

I was looking at the cost thing too. I used a super cheap service once for some small parts, and the big problem wasn't how fast they arrived. When the boards got to me, the little slots where the wires connect were just a bit too wide, and they wouldn't sit right in the plastic box I made. It seemed like the local shops might catch those physical fit problems better before they even make the board. It feels like the whole thing is less about the electronic design and more about making sure the physical stuff matches up. Is it true that quality is always going to be diferent between the cheap shops and the more expensive ones?