I'm just noticing this post today, so I can't provide any feedback about the actual experiment.
I do think that if the half index is as 'bad' as it seems from some replies, I would not be using Kagi in that case any more. The quality of the results is the #1 benefit for me. All the rest (like no Google) is also nice and something I want, but that doesn't justify the costs (then I could also use another free Google alternative).
Also, I'm not sure if offering the half index as a separate price would appeal to me. As paying customer, I would like Kagi to be focussed on providing the best search results possible (as that is my #1 priority). If you also offer a 'half index' plan, your business is clearly not fully focussed on that goal, and that might reduce my sympathy for Kagi.
If the current price is not sustainable with the full index, I would prefer a $15/mo option (for the full index), and no 'lower' option (guessing/hoping that it might even out).
If you want to do any more experiments like this in the future (or if you really are going to offer the half index option), maybe it would be interesting to make an interactive version of it? I could imagine that you could see the results (half vs full) side by side, you can can more easily (and in a justified way) compare it. Or make the 'extra' results from the full index half-transparent, so you can clearly see what could be missing (though that might not work if the ordering of results is also different with the half index).