I'm going to assume you want to make the best applications you can on each platform and are not willing to accept a lowest common denominator solution that is sub-optimal everywhere.
The only good option is to separate your UI and logic and write a new UI layer for each platform. It is hard work and there's no easy out. On a phone, the UI has to be touch only. On Windows you probably need to support touch and keyboard/mouse. Linux may lead you to develop a TUI and GUI. Your macOS version should probably be a GUI.
Then on each platform, there are significant differences for things like where the settings UI should be and where the settings should be persisted.
Your case here feels wrong. The goal is shipping something that provides value and can have features added to it at a pace customers expect, fixing bugs quickly, etc. Having a separate UI layer and needing domain knowledge for every single platform, coordinating releases accross all these different platforms, and everything else that comes with a cross platform app is quite simply not worth it from a business perspective. Especially with the idea that if you spend half the effort you would from developing 4 native UIs on wrangling the performance hit from Electron to maybe only 20% worse than native, it' just not even a consideration anymore.
It comes down to choosing quality or quantity. You can't have both. Obviously lots of people think like you and choose a lowest common denominator solution.
And you are seriously underestimating the cost of Electron (which isn't an option on some relatively big platforms like Apple Watch). Memory consumption, CPU utilization, and battery consumption are the key factors and a well written native app will always beat a well written Electron app usually by more than 20%, especially on battery consumption. If you are lucky and have millions of users, then multiply that across all the users and it's a tremendous amount of waste.
I can see why you might choose Electron if you have very few users or the app is essentially a web page.