Email dedup has three knobs: policy, scope, and BCC handling. Defaults work for most matters. Last reply plus unique catches forked threads with unique attachments. Global collapses cross-custodian copies. Default (ignore BCC) groups sender and BCC copies as one. Switch defaults when the case calls for it. Disclose your choices to opposing counsel in writing.
Options and fields
Policy
Policy decides which copies count as duplicates inside a thread. Two choices: Last reply only and Last reply plus unique (default).
Worked example. A five-email thread. Robin sends a contract draft. Jordan replies confirming section 4. Casey replies with a redline attached (unique attachment). Robin acknowledges. Robin sends v4 with everything quoted. Which copies stay?
Last reply only
Only the latest reply in the thread is kept. Everything earlier is hidden, including the mid-thread message with the unique attachment.Last reply plus unique
The latest reply is kept, and so is any earlier message with a unique attachment or unique content not subsumed by later replies.Last reply onlykeeps just email 5. Casey's redline attachment vanishes from review because email 5 quotes the body text but doesn't carry the attached file. Fast, aggressive, lossy. Last reply plus uniquekeeps email 3 and email 5. Casey's redline survives because it's a unique artifact. Use this default unless you've negotiated a last-only protocol with opposing counsel. The Sedona Conference and EDRM Processing Standards align with the last-plus-unique pattern for inclusive-only review.
Scope
Scope decides whether dedup runs across the whole case or per custodian. The two choices map to two case shapes.
| Aspect | Global | Per custodian |
|---|---|---|
| How dedup works | Runs across the whole case at once. A thread is collapsed once, no matter how many custodians' mailboxes it landed in. | Runs independently per custodian. Each custodian's copy of a thread is dedup'd in isolation. |
| Reduction ratio | Higher. Cross-custodian copies collapse. | Lower. Each custodian retains their own thread copy. |
| Custodian provenance | Captured on the inclusive copy as a list of all custodians whose copy was subsumed. | Preserved per custodian. Each custodian's thread sequence stays distinct. |
| Use when | Multi-party litigation where the conversation matters more than who's holding which copy. | Internal investigations where each custodian's exact sequence matters (who saw what, when). |
Use Global for multi-party litigation. The production set is what matters, not who happened to hold which copy. Use Per custodianfor internal investigations where each custodian's exact sequence matters. If you're building a timeline of who saw what and when, you don't want one custodian's inclusive copy absorbing everyone else's.
BCC handling
BCC handling decides whether differences in the BCC list split a dedup group. Two choices.
| Aspect | Default (ignore BCC) | Strict (include BCC) |
|---|---|---|
| What counts as a duplicate | Sender's sent copy and the BCC recipient's received copy collapse together. BCC is invisible to dedup. | Sender's copy and BCC recipient's copy stay distinct. BCC differences split the dedup group. |
| Reduction ratio | Higher. Fewer dedup groups. | Lower. More dedup groups retained. |
| Trade-off | You see one copy of an email that may have had multiple BCC routes. Saves review time. | You see each BCC-distinct copy. Useful when establishing who actually received the message matters for the case. |
| Use when | Most matters. BCC rarely changes the substantive content that matters for review. | Investigations where the BCC list is itself probative (for example, secretly informing a third party). |
Default for most matters. Switch to Strict when the BCC list is itself evidence (someone secretly informing a third party, for instance). The trade-off is a lower reduction ratio and more emails to review.
Defensibility
Email thread dedup is widely used but should be disclosed to opposing counsel and noted in the ESI protocol. Courts have upheld threading when parties agree to it upfront in writing. Courts have ordered full production of non-inclusive emails when parties did not negotiate the practice first.
The Sedona Conference Cooperation Proclamation and the EDRM Processing Standards both recommend documenting dedup choices in the ESI protocol before collection ends. In re Tecfidera and In re Actos are the two cases practitioners cite most often: both turned on whether the producing party disclosed the dedup practice in advance.
Always disclose your dedup choices in writing before production. Pin the policy, scope, and BCC mode in the ESI protocol and keep the proportionality argument ready: thread dedup reduces review burden without losing evidence. The hidden copies stay in the case record and remain available for production-side reconciliation if challenged.
Edge cases
- A forwarded thread with edited quoted text counts as unique content under Last reply plus unique. Hintyr keeps that copy.
- Mid-thread attachments not re-attached in later replies are unique. The policy default keeps the message that carries them.
- A thread split into a forward chain and a reply chain is two inclusive copies. Both survive Last reply plus unique.
- Per custodian scope doesn't collapse cross-custodian duplicates. If two custodians both received the same external email, you'll see both copies. Use Global if that's what you want.