I don’t believe there is a one size fits all split, it depends entirely on the type of extension and what value that extension adds to the stream.
For example I have a monetized extension that I’d be fine with a 0% cut of bits, because I was hired to make it so have already been paid for my work. I’ve also had charities reach out to me as I work some charity broadcasts who were interested in extensions, now Twitch flat out doesn’t allow the sort of charity extension so I couldn’t make them, but if I could then a 0% split would also be appropriate there.
There are also some extensions where something like a 5% split would still cover operational costs and generate some profit.
To give an example why 20% can actually stifle the Bits ecosystem, in the UK 25,000 bits is £299.56, or $400.78. And that’s with the max 12% discount on bits so is best case scenario. If those bits were used in an extension the broadcaster would get 20,000 bits as their 80% which is $200. So for a viewer spending $400.78, the broadcaster gets less than half of that. Worst case, buying smaller amounts of bits and no discount, 25,000 bits would cost the viewer $455.11.
Note: Bits may not be quite so bad where they aren’t taxed on purchasing bits, but it’s still not great even in the US where its cheaper to buy bits. Plus this also shows that it’s less economical to use bits in some regions, so extensions with monetization will less reliably reach all international broadcasters.
If the viewer was to cheer those bits instead of using the extension, the broadcaster would get $250.
So you have to think, what value does your extension add that is worth the broadcaster receiving half of what they would have if they simply linked the user to their PayPal, or other 3rd party payment processor?