Generally we do not want to include PII information in the data feed, but it may be necessary. For example, in collecting data for analysis or aggregation, we would at least need the usage point (address).
Today, PII information would be inserted into title fields or other fields. The problem then of course is it is difficult in identifying PII data since it could be anywhere - which makes anonymizing the data very difficult.
I propose adding a new, optional element to the schema so if someone did need to include PII information, there was a specific place for it. And since there is a specific place for it, then it would be easy to remove as a step in anonymizing the data.
Optional fields that come to mind are:
usage point street address, city, state, zip, census block (as separate fields of course)
account owner name (eg someone is downloading their data and we might want to have their name on it so they know for sure it's their data).
and perhaps we could have generic attribute/value pair elements to handle other things (meter type, a special experimental cell they are part of, etc). So if there is additional meta data that is needed for a unique purpose, there would be a specific place in the XML for it to go.