Well, let's say we were designing a budget oscilloscope and our market research found that suddenly absolutely no one wants to buy anything with a noisy fan anymore (although for some reason most of the survey participants had the same IP address based in Italy, weird that). So we decide to make silent operation a prime design goal and "go to town". Good thermal design is free after all, right? So what do we do?
- massive custom heatsinks help; they are also expensive and require mechanical support
- bigger, slower fans are a lot quieter; and you will probably need to design a new, bigger case to fit them
- well, if we are designing a new case and have new injection moulds made, at least we can re-design for better airflow as well (maybe hire a specialist?)
- oh no, the PSU has minimum airflow requirements we don't meet anymore; well, not to despair, we just get a more expensive one with higher-rated parts; or, how about a custom one that integrates nicely with our new case, I mean how much can that really cost?
- our main board may now run a bit hotter, that may be bad for realibility. Oh well, let's order some Nichicon 105°C caps and use the Crapxxon stockpile for something else instead.
- we also have a potential for larger temperature-gradients in the frontend now; may need to loosen some of the specs a bit? Or find some lower TC parts?
- the old model had a lot of headroom when it comes to cooling really, we figured "better safe than sorry" and were fairly confident it would be able to take almost anything customers would throw at it even with the vents all clogged up after a couple of years; we're flying a lot closer to the sun now but we'll just do really thorough environmental testing, so we will be fine
Fast forward to the launch date... Well, it was tough, but we made it.

Ok, kinda. There were a lot of delays because of all the custom parts and the re-designs. Also the bike-shedding in design review meetings, everyone is a bloody expert in thermal design now, it was basically like reddit in there, soo annoying. And there was that embarassing incident when a prototype set a customer's lab on fire. But the software people say the fan-control bug is fixed now.

So some minor features didn't get done in time and we moved most of the devs over to the new spectrum analyzer project. We just tell customers that "there is no public roadmap for new features at this time but software development is ongoing". It's not a lie either, Karen comes in twice a week to fix some bugs. Anyway, it's mostly just statistics and some weird memory-management stuff that no-one cares about that's missing. And it's just twice the price of our direct competitor's product, which is soo much louder.
It's a great day, it's just a little sad that our social media manager quit today because he "couldn't take the abuse anymore". Some guy calling himself balnazzar was spamming rants everywhere about how the new scope "has really bad price/performance" and is "useless because it doesn't have basic features" and that he would rather use a rusty nail and a light bulb.
* I originally wanted to write a serious post about thermal design but I am currently home sick and may not quite be my normal boring self. There is stilll a serious point here: nothing is free, there are almost always trade-offs.