Your better bet would probably going the PiStorm route, which is essentially a combination of Raspberry Pi and a simple CPLD with the former one doing the actual heavylifting. In the current state, only regular 68000 DIP packaged can be replaced by it. Even if it's a project for Amigas, someone already has reported success in Macs:
/edit: Just as a sidenote, the PiStorm combined with a RPI 3A+ yields about 50x the performance of a 25 Mhz 68040 and 1700 times the performance of a regular 68k ....
Someone asked this guy what it benchmarked at and he said 1.7x stock.
Some of the Amiga upgrades that drop in the CPU socket have been proposed for mac. I got a little frustrated because some people (I suspect not even the dev) were encouraging Mac people on facebook to fund the crowd funding thingy. They were talking about all the RAM and hundreds of % speedup. I belive that some compatibility is possible, but without likely re-writing parts of the ROM, making low level OS patches and and reigning in the expectations, this is going to take serious work. I feel that it was so unlikely to happen (I really hope it does!) that taking these people's money, or at least persuading people that are unlikely to benefit to fund a project they wanted to happen, saying it would give you a 512MB Mac Plus was... immoral.
As it stands, dropping in a faster 68000 replacement in a compact mac is going to cause the issues that the old upgrades struggled with back in the day - unlike the Amiga (I might be wrong - I've never owned one), the mac uses the CPU to control some timing critical hardware while turning off interrupts. These include the floppy drives (the cpu has control, directly, over spindle speed and head movement I believe), serial ports and sound at least. I don't know of a 68000 era mac processor upgrade that doesn't mess up sound!
This means that to get one working we need to catch ROM calls to functions related to these hardware devices and carefully make sure everything happens at the right times, likely slowing to stock speeds while doing it. The people with these skills are few and far between in the MacWorld sadly. The professionals seem to have moved on and even some of the documentation is hard to find. Another challenge with RAM is the 68000 Macs use the motherboard RAM as a framebuffer (two?) and the video circuit uses DMA. This would need to be considered. Also, the stock ROM (except the Portable?) is in the address map straight after the RAM, and you need continuous RAM on these machines. An MMU could solve this, but again, I don't have a clue where you'd start with that.
Getting a 68060 working is probably closer to possible, but nobody has one working that I know of. That 630 comment is really interesting - absolutely worth having this discussion just to learn that. I'd like to have a go one day, but might never learn enough to know how, be smart enough, or find the time.
I suspect all three. Plus my cat keeps sitting on me and I feel guilty pushing them off to do project work.
I imaging once you'd gone through all manner of horrors patching the ROM, you'd set up traps to capture the unimplemented Opcodes and load them in an extension as soon as possible. Then it would effectively live patch any software. Hopefully the OS doesn't call any early on or... more pain.
Since we're talking about it, what I want to see is someone bodge a PPC750 onto a Daystar 601
Its all 601 bus! 100MHz G3 in a IIci?! An inefficient waste of time? I didn't realise you knew me so well
Although thankfully I'm not crippled by a bus...
Lastly, we've probably been a little too negative and I just want to say, if you start trying to implement the mods needed, I may not be much use but I would do everything I can to help. I suspect progress would be the biggest motivator. As things are, I don't expect to see it happen anytime soon.