But the eight million dollar project is more efficient than waiting for a random twenty dollar fix. And as it seemed to be meeting the goals set for it, it was also likely more efficient than allowing bad product out the door, which is what this is really gauged against.

Honestly, I was expecting the punch line to be that they found it cost more in the end because it shut the blasted line down for each bad box. I'm an engineer in a plant, and frankly that just seems insane, even if localized (accumulators are never big enough!)

I think it might have been more efficient to hold a company meeting that said, "Hey guys, we are having a problem with empty boxes getting shipped. If you have an idea of how to make sure empty boxes get removed from the line, let us know!" Note that the simplest and most cost effective solution will be used, and offer a thousand dollar bonus to whoever suggests it. (The bonus might motivate people who otherwise might not care.)

That is orders of magnitude more efficient than commissioning an $8M consultation, and is likely to find any low-hanging fruit such as fans-on-conveyor belts. If no one has any ideas in the first month, it's possible that such insights are not commonplace, and then you can go commission your million dollar consultation.

I'm going to pretend that you didn't suggest holding a plant-wide meeting as a form of efficiency ;)

But the idea of setting up bounties is controversial. The arguments for and against it quickly get remarkably tortuous. I've never seen it implemented, but I can definitely say the idea is attractive. I had a coworker who would offer a hundred bucks (of his own money) to the operators who maintained stable production on a weekend run; it looks like it works, but he couldn't afford that too much.

It's often easier to spend giant sums of cash than to be clever, as cleverness is unreliable, but giant stacks of cash are accountable. (TL;DR: industry is often penny wise, pound foolish.)

I wouldn't call a solution gathered without consulting everyone's feedback 'efficient'. Obviously the 'fan solution' might have come up earlier (or at least some derivative)