Last updated:
Why an e-discovery platform, and not filtering on site
Sooner or later a party asks: why did you take the complete mailbox, if the assignment was only about a limited period and a few keywords? The answer lies in the digital trail. Even months later you have to be able to show that you took nothing more and nothing less than needed.
How you do that depends on where the mailbox lives.
Mailbox in Microsoft 365: filter on the cloud side
If a small company has its mailbox in Microsoft 365, Purview is built to filter on the cloud side. Taking the complete mailbox and then filtering it locally cannot be justified there. It is simply not efficient either. Work in a targeted way and filter on the email addresses and search terms in the court order. How is explained under Microsoft Purview eDiscovery.
Mailbox on a local server: everything first, then filter
If the mailbox is on a local server and you can only consult it with Outlook, filtering on site does not work, and through the cloud it does not either. Then you take everything first and filter later, for example in GoldFynch. The same goes for a company drive.
Outlook does not export a selection
In Outlook you can search and look at the hits, but you can only export per complete mailbox, not per selection. If you still want to select on site, you have to move the emails found into a separate folder. That changes the custodian’s mailbox.
Searching needs an index
Whoever opens a copied mailbox in Outlook has to wait until the search index has been rebuilt. For a large mailbox that takes hours, and how many items are still waiting you only see under Search Tools > Indexing Status.
Searching on an incomplete index simply returns results. Outlook only puts a narrow bar above them saying the results may be incomplete (Microsoft Support). Whoever misses it decides on half a dataset.
Searching content depends on the format
With what is on an ordinary computer, you cannot search the content of PDF, Word, PowerPoint and text files in one go. Each format needs its own text extraction. If it is missing, the computer silently searches only file names and metadata. A company drive with all those formats mixed together cannot be searched completely on content that way, and opening each file separately is not feasible with a few tens of thousands of files.
Filtering afterwards can be checked
Whoever selects on site decides under time pressure at somebody else’s desk, and those decisions are recorded nowhere. Afterwards it can no longer be traced why one file was selected and another not.
In an e-discovery platform every filter step is a separate action, with a count before and a count after. An opposing expert who takes the same steps arrives at the same counts.
You throw the bycatch back
If you take more than requested, you vouch for it on your honour as a sworn expert. Like a fisherman, you throw the bycatch back into the sea:
- You deliver nothing that falls outside the assignment. When in doubt you do not decide yourself, but put the question to whoever gave the assignment.
- You delete the source data permanently afterwards, from the platform and from your own media, and you note in the report when, where and how.
Further reading
Craig Ball, Don’t BE a Tool, GET a Tool! (22 March 2021). Ball reaches the same conclusion by a different route: looking at evidence in the program that normally opens it changes the metadata and with it the hash value.