The chronological or role-based approach
Some companies adopt a chronological approach to determining which e-mail messages to save. For example, they might provide users with simple retention "buckets", such as mail folders labelled "Three Years,""Five Years" and "Seven Years," and then instruct users to place e-mail messages into those folders accordingly. Other companies consider a role-based strategy; they might capture and save e-mail from IT staff for one year, e-mail from sales staff for three years and senior management's e-mail for seven years.
Both approaches share the same fault - the e-mail retention policy is not based on the content of the actual messages. In other words, e-mail is being saved simply because it is e-mail. Companies who adopt such practices typically find that employees do not think in terms of a simple time-based retention; instead, people in their daily jobs, such as a decision or they think in terms of what messages actually represent. So, not surprisingly, most users will simply place everything into the folder with the longest retention since that is the safest and easiest decision. Similarly, basing e-mail retention solely on a user's role within the company is flawed because users often have many different roles. This creates dotted lines that complicate not just their reporting structure, but also the information with which they work.
The content-based approach
Other companies adopt a content-based strategy. For example, matter-related messages are identified as such and inherit a retention period appropriate for content of this type - perhaps being saved for 10 years following the completion of the matter. A content-based strategy is typically optimal because it gives organisations a better sense of what information has actually been retained and more granular control over ensuring content is kept and disposed of in a structured fashion.
Some organisations pursue a strategy characterised by users making overt decisions to identify the e-mail messages that should be kept, meaning messages are transparently captured into a managed environment and retained in accordance with policy.
Other organisations investigate possible options for an intelligent system to evaluate content and make a determination on behalf of the user.
It is plausible that an intelligent engine could, perhaps, identify and capture some intended record messages, because such e-mail messages might have a contract attached to them, and an intelligent system could compare the contract to a "sample contract" that had been used to train the system. Or outgoing e-mail messages might have a client or matter number inserted into the subject line, which the system could examine to trigger classification.
Consider that e-mail is notoriously contextual and often deliberately vague. What has business value to people in their daily jobs, such as a decision or approval, could contain just a few words, making it very difficult for an automated system to identify as useful to keep.
Sign up for Computerworld eNewsletters.