Customer Data Backup Policy
Last Updated 1/22/2018
Customer Science classifies data into two broad categories - 'atomic data' and 'pattern data'. The atomic data is the raw data to which customers give us access. Pattern data is the patterns our AI finds in the atomic data, as well as additional information that our customers may add to it (ex. a comment on a pattern).
Customer Science deliberately does not make backups of atomic data. We attempt to actually "self-destruct" all such data where possible as that data has the highest sensitivies. Thus we preferentially destroy cached versions as possible and rely on customer data stores as much as reasonably possible. In the event of data loss on our side, we will expect this data to be re-populated from the customer.
Pattern Data is aggregated to such a level that it is less security sensitive (though clearly we treat all data as sensitive). It is backed up in accordance with normal SaaS processes. This includes the use of redundant databases where reasonable, and daily backups with more frequent snapshots. In the event of major data loss, we will restore from the previous daily backup or the most recent snapshot if possible.