Here's an idea or two.
Wireguard. Stop sitting on your hands complaining about how wireguard isn't mature, and support it with the generic native apps (now there's even a (beta) windows client). The network address selection issue requires engineering effort, but wireguard itself is most likely not going to address that soon, because it's designed to be a minimal vpn codebase, so why don't you engineer a solution yourself? Or use NAT like nordvpn apparently does.
Explicit stock OpenVPN support. You kind of do this, but it's still difficult or off-putting for non-technical users to figure out which config to grab and how to install the stock client. On your setup page, make sure you're providing a link to the stock (windows) openvpn client and install instructions for Mac and major linux distros, so that people who don't trust your binary blob installer can use the generic one (minus all the fancy stuff like pretty config for auto-selection of endpoint, showing port number, DNS and kill switch things). Make sure to provide sample configs that are up to date and usable.
Nobody has to trust your software if you make it easy to use a generic client instead.
I realize Wireguard is tricky because it doesn't have ephemeral net address selection built into the protocol, but can you please just get that support done? What is your dev team doing if they're not doing that? They don't have to maintain openvpn, unless continually tweaking the custom UI is their prime focus. I'm tired of OpenVPN's instability and risk from its gigantic codebase. I don't care if wireguard has lurking bugs that make it insecure against the NSA. The NSA is not my threat model. You can support wireguard while cautioning everyone that you don't trust it as much as openvpn, and then let them make the choice based on how much they trust you, how much they trust wireguard, and how much they trust Matthew Green's audit of openvpn.