Several on here have mentioned doing proper market research to prevent overbuilding of product. If that worked you wouldn't have such things as TP shortages or bouncing prices of petroleum products.
Note that these are not quite the same things! As I recall, TP producers never ran out, and shipments from warehouses were always prompt, it was just local (daily) stocking running out (but, I don't know the actual numbers of failure-to-deliver, if any). Reflecting this, the price hardly changed AFAIK.
Petroleum prices were all over, last year, and that came with some interesting differences. It's liquid energy, a little too precious to dispose of (or, fortunately for us, illegal to?). Every barrel of spare capacity was used up (full tankers, anchored anywhere they could), and at some points prices even went negative ("please, for the love of god, TAKE my oil!"). This reflects the knock-on costs of stopping and restarting production (many production systems can't simply be throttled down, and take huge effort to cycle).
(Likewise, the current semiconductor shortage, is in part the consequence of following through on the command to shut down production lines. If not entirely, then at least the orders that were cancelled prematurely (and impulsively so, it turns out). And it's taking a long time for those orders to filter through and resume.)
It's actually an interesting pair of examples to contrast. In one case, production or reserves were ample and prices remained stable; in the other, production was stable but demand went down suddenly, saturating reserves, and sharply reducing the price. In an earlier time, perhaps the excess production would've indeed been dumped or flared; the Cuyahoga river didn't just catch fire once (though as I understand it, that was more due to industrial runoff/waste, but one could imagine such a direct route as this being employed, given the prices in question..).
Somewhere between these extremes, is likely where common electronic products would land. (Which is what we're talking about, right?) Some are fairly trivial to build, just order more from the CM and they're there in a few weeks. Does assume everything in stock or quick to make (machined/fabricated enclosures may take a few more weeks, new moldings maybe a few more still?), YMMV. Stock could be tiered with some extra capacity at the business (or getting the CM to hold some, at whatever rate they charge) to minimize excess that Amazon carries directly. Only ramp up once you have established sales, don't drop a hundred and have to burn almost all of them because the launch failed. Establish a meaningful (efficient, responsible) RMA program, so returns aren't just shredded. Etc.
The relative amounts for all of these steps, varies. Few steps might be applicable to commodity crap that's barely worth its weight in plastic; pricey stuff like laptops, Apple products especially, test equipment, etc. is surely worth returning to supplier, they just need the programs in place to handle that action (and, if they think such programs cut into their profitability (which Apple apparently does, among others), then the externalities would need to be, in effect, applied to them via appropriate laws).
Bottom line, this waste is the lubricant that allows our society to perform at the level it does. While some tweaking of the process can and should occur there are only two solutions - accept waste or dramatically reduce the material standards we have become accustomed to. Well, there is a third, dramatically reducing world population, but there is no way for that to happen in a comfortable, equable way.
How are there only two binary solutions, and not a continuum of compromise, depending on product type / market / scale / etc.?
The third option seems to be, reduce the profit margin standards. Incorporate externalities into business expenses. Not that you'll get anyone to support such a thing, certainly not in the current [neoliberal] climate. But that's the most obvious route to me.
Speaking of population, it seems human birth rates decline as standard of living, and cost of living, rise. Eliminating excess infant mortality, seems to be a particularly powerful driver.
If that counts as "killing them with kindness", I suppose it does. But most moral frameworks find it's an acceptable mechanism, to reduce a population by beneficial and voluntary actions. Put another way: taking a life is immoral, but deciding not to create new life is moral.
Which I think is interesting by itself, as common moral frameworks are built around the individual, or groups thereof; up to scales of whole societies perhaps (e.g. rule utilitarianism), but still within a horizon of individual lifetimes. Traditional moral systems don't seem to say much at all about behavior of the superorganism, over evolutionary, let alone geologic, time scales (1ka+). Such is still very much in the domain of raw evolutionary realism: the entity continues to exist, or it ceases to, and that's that; no cosmic power cares one way or the other. (Well, beyond the inevitable march of entropy, which stands to "benefit" from the catalytic effect of life as we know it.)
Tim