Probe brainstorming:

Web password use:
* # of distinct domains with password forms submitted in a session

Password Management Probes:
* number of saved passwords
* number of distinct passwords / number of passwords
* average/median password age
* password use counts (make sure they work first)
* number of origins the password manager is disabled for
* flag whether pw-saving is on/off globally
* telemetry on whether passwords match one of the 1000 most common

Form fill/save accuracy
* % form submissions for an origin where the submitted combo doesn’t match a saved username/password/both
** i.e. % of form submissions where we have the submitted info already saved
** don’t double-count same origin?
* record in histogram for each of the reasons why we don’t autofill or save (e.g. autocomplete=off, multiple signons available, etc.)
* count corrections to which field is auto-filled (e.g. if we guess un/pw field wrong)