Parallelized service start on desktop boot and other similar desktop boot optimizations come to mind.
But yeah, mostly it's the remaining 80% work of polish (in the sense of the adage of the remaining 20% actually being 80%) needed to make a Free Software system usable by the masses. Lots of work and lots of value but this doesn't involve significant technical advances, just hard work by smart people who truly understand the whole stack technically but are also able to see from the perspective of ordinary users.
Canonical also pioneered Launchpad and Bazaar. Today they are eclipsed by GitHub and git, but at the time they led in Free Software distributed development tooling. Early GitHub was apparently modelled on Launchpad for example. There's a really interesting retrospective that explains a lot of what you see of Canonical and community and contributions here: https://www.jelmer.uk/pages/bzr-a-retrospective.html
Oh, and container technology. You'll need to look into the details of that to understand exactly what Canonical did and didn't contribute, but Canonical's contribution was significant.
> NIH
If you look into the specifics most of the things commonly cited as Canonical NIH predate or at least were already in progress at the time of the alternative.
Disclosure: I work for Canonical but wasn't involved in the work cited above.
This was a pretty big deal at the time before broadband was easily available.
Canonical was also focused on user friendliness more than any other distro. Which was also important for desktop Linux’s growth.
I must be missing something, but TBH I’m finding it hard to remember any major technical contributions from Canonical.
They’ve also had a pretty disturbing history of NIH syndrome, wasting years of effort on alternatives to what the rest of the Linux community is going forward with before deciding to abandon their alternative.