I don't know if you have tried this yet, but sometimes having a lookalike custom audience that is 99% similar doesn't always preform better than the lookalike audience that is 97% similar.

If I were you I would go through AB test each of the different lookalike audiences, so try 99% vs 98% vs 97% vs 96% vs 95% and see if the less similar audiences will still give you good results, because they should also have a larger audience group for you to market towards.

Also have you AB tested interest groups? I'm not sure what your exact niche is but i've worked on campaigns where certain interest groups preformed better than the lookalike audiences.

If you're spending a decent amount of money on Facebook ads, having a professional go through and AB test your ads can be a very good idea. It's the difference between having a good ad & knowing you have a perfect ad.