Make sure you can defend your adjustments.
Exactly. And you also have to be able to defend an adjustment of zero.
In this market, clients are increasingly skeptical of zero.
In my analysis, I'd start by citing the total MLS report's overall statistic, add a comment about the MLS statistic for the sub market or neighborhood, and then determine the statistic for the handfull of sales that are most comparable.
The total MLS statistic in my market (and considering the local specific quirks in my local MLS) is accurate enough and a large enough sample to be credible. Of course that statistic is always a ratio to the "final list" and does not consider listing history.
Depending on my submarket size, it may also be accurate enough and a large enough sample to be meaningful. If not, I tweak it or explain why it is meaningless.
My short list of truly comparable sales can also be queried for automatic MLS statistics and "purified" if needed.
Then I have all three elements to reconcile. I might give more or less weight to any of the three elements depending on what I know about the data.
Once you have reconciled the data, it would only take one or two sentences to report your reasoning in comments.
=======
Lender clients usually fall into three camps. Camp 1 will question you if your adjustment ratio is not directly demonstrated by the comps used in the grid. Camp 2 will question you if your adjustment ratio does not match the greater market. Camp 3 will never question that adjustment.
As a usual practice, one needs to be prepared to back up their reasoning for both camp1 and camp 2.