Impact case studies must be based on research that is of quality 2* or greater. In many ways this should be an easy target, if research is sufficient to create impact it is hard to imagine it not being 2* quality.
However, there were cases during REF2014 there the papers used as case study evidence were also submitted as outputs for scoring. In some this was unproblematic, but there were a number of cases where the output scoring had been 1*. In these case there was no room for generosity, the case study failed the ‘threshold condition’ and hence risked being completely zero rated. In most cases this affected only part of the case study, but even so a substantial part of the impact in the case study was voided.
Oddly it seemed clear that in some of these cases the staff member would have had stronger outputs and it seemed likely the impact case study output was only submitted in order to let it be scored 2*. Of course if it failed to get 2* not only did this weaken the output scores, but also brought down the case study.
I call the double jeopardy!
Unless the rules change substantially for REF2021, do not do this! It is a gamble. Instead, if you have any doubt do not submit case study evidence papers to the output scoring process. If the paper has been published in a reasonable international venue, it would be a cruel Case Study assessor who did not rate it “quality that is recognised internationally”. However, if it is in the outputs assessment process there are no guarantees.