Question 1

Does sharing our search terms reveal our case theory?

Accepted Answer

Not if the negotiated terms are built as they should be: from the issues and the estate's vocabulary, which any competent opponent already derives from the pleadings. Theory-driven searches (the ones chasing your private hypotheses) are a different artefact: run them internally across your own review population as work product, and never put them in the negotiated set. The negotiated list culls; your private searches find; only the first is exchanged.

Question 2

Must we exchange hit reports?

Accepted Answer

No rule compels it in terms, but the regime's logic does: the DRD asks for search proposals the court can assess, PD 31B directs discussion of keyword strategy, and a term proposal without numbers is an assertion the court will discount against a counter-proposal with them. Practically, the party with hit reports wins term disputes; exchanging them is both the cooperative act and the winning move, which is why they are prepared exchange- clean from the start.

Question 3

Can the other side see our TAR training set or document scores?

Accepted Answer

The established line: methodology and validation statistics are exchangeable process (workflow, sample designs, richness and elusion rates); the model's contents (training documents, per-document scores, ranking order) map onto your team's reading and prioritisation and are work product. State the distinction early and hold it; offer statistics generously, and the demand for the model's insides usually evaporates, because the statistics were the legitimate part of the ask.

Question 4

What if the other side simply will not reciprocate?

Accepted Answer

Perform your half visibly and completely, request theirs in writing with deadlines, and let the file work: at the CMC or a Disclosure Guidance Hearing, a party with an evidenced package facing a party with adjectives obtains its scheme and usually its costs. Do not descend to symmetry ("they withheld, so we will"): the conduct ledger only protects the party whose side of it is clean.

Question 5

How much privilege-log detail does cooperation require?

Accepted Answer

A proportionate description that lets the other side and the court understand the claim's basis: commonly category-level for routine classes (advice correspondence with named firms in a period) and item-level only where a claim is genuinely contestable. The log's structure is negotiable process; the content it protects is not, and no cooperation duty requires descriptions so granular that they erode the privilege they exist to record.

Question 6

Is agreeing a clawback arrangement a sign of weakness or sloppiness?

Accepted Answer

Neither: it is standard engineering for volume productions, protects both sides identically, and supplements rather than replaces careful privilege QC. In TAR-scale exercises the realistic error rate is never zero; the agreement converts the inevitable rare mistake from a waiver battle into an administrative step, which is exactly the proportionality the regime exists to promote. Proposing it early is read by courts as competence, not concern. cflab. u k · e-disc ove r y. u k ©2026 Computer Forensics Lab Ltd ·cflab.uk ·e-discovery.uk ·info@cflab.uk ·+44 (0)20 7164 6915 Page 19 of 22

Guide 17: Cooperating on Electronic Disclosure Without Giving Away Your Case