> screen reader users was left with a sub-optimal experience ... but hey they get to use the same URL as everyone else so there is nothing for them to complain about.
Sure, if the website is sub-optimal, the users are free to complain about it (and file a lawsuit, depending on the country and laws). From my experience, UX and Accessibility go hand in hand. So if your website is sub-optimal for screen reader users, I can almost guarantee it will be sub-optimal for other people as well.
In theory, this is a false dilemma - it's not either a highly optimal version for screen reader users vs. a sub-optimal experience with the default version. The third option is to "simply" make the default version accessible (which indeed may mean _a lot_ of work = money, especially for legacy projects).
It's like creating a separate version of your website just for mobile devices. This pattern was used a lot in the past, but not that much anymore. If you build your website with mobile and accessibility in mind (mobile-first, a11y-first), you get an overall better experience.
> We worked together with Jaws to find a way to determine if the user was using a screen reader.
This is basically the opposite of what I just wrote. There was a big uproar regarding the new Apple feature [0] for doing exactly that.
> Dont know if things has changed since.
My personal anectode: I was part of a complete re-write of a big airline website and we followed the mobile- and a11y-first approach with much success. I think the major part of our success was having _everyone_ on board from day one: designers, developers, testers, product managers and customer had all been working with accessibility in mind.
[0] https://axesslab.com/digital-apartheid/