I don't think we will be able to support bucketing twice but allowing you to define number of docs per bucket instead of number of buckets can help in the granularity issue you spoke about. For example, if you choose 20 buckets, an item ranked 50th by text_match in a query that produces 1000 hit might be boosted to the top, while an item in a 20-result case might receive no boost at all. On the other hand, if you define a number of docs per bucket as 20, you should get adequate boosting for records with similar text_match scores in the 1000-results case. In the 20-results case, you would still see the most popular content first. This is unlike the number of buckets configuration.