Senior Counsel
January 20, 2026
Large-scale production orders are no longer exceptional. What is surprising is how often sophisticated litigants underestimate what courts now expect when those orders arrive. The Federal Court’s recent decision in Commissioner of Competition v Amazon.com.ca ULC, 2025 FC 1782, arises in a competition-law investigation under the Competition Act. But it would be a mistake to read it as confined to that regulatory context. The Court’s reasoning reflects expectations that increasingly apply across complex civil litigation, particularly cases involving large-scale electronic production. In substance, this is a decision about modern litigation discipline.
Ten practical lessons stand out.
First, preparation must begin before the order is made. The Court was clear that experienced counsel should know that draft production orders are rarely expanded and often issued substantially as proposed. Waiting for the “final” order before identifying custodians or mapping data sources was treated as a strategic choice, not an excuse. That logic applies equally in major civil cases where counsel have early visibility into the likely scope of production.
Second, 120 days (where mandated) is not a starting point for negotiation: it seems now to be the outside limit, even for very large matters. Although the Court acknowledged the exceptional breadth of the request, it still treated 90-120 days as presumptively reasonable. Extension requests approaching eight months were viewed as extraordinary and requiring compelling justification. Civil litigators should expect similar skepticism when timelines drift far beyond established norms.
Third, custodian identification must happen early and must be actively managed. The Court was particularly critical of delays in identifying custodians, noting how rapidly the list expanded once focused attention was applied. That contrast undermined claims that earlier delays were unavoidable. Custodian work is a foundational step in any serious production effort, and any delay at this stage threatens the entire production process.
Fourth, rolling productions are the default where large data volumes are concerned. Amazon emphasized their efforts at rolling productions, but the Court treated that as table stakes rather than a concession. Whether in regulatory proceedings or civil discovery, courts increasingly expect requesting parties to receive usable material early so cases can progress while later tranches are still under review. This is especially important where previously reviewed materials are available to the producing party at the outset.
Fifth, courts are openly questioning a rigid (and in our view outdated) two-pass review model once technology assisted review is deployed. The reasons scrutinize whether a traditional American-style 1L/2L workflow remained proportionate after analytics had already reduced the dataset. This signals a broader willingness to question legacy review practices in civil litigation, particularly where technology is invoked to justify delay.
Sixth, even where the legacy models are accepted, standard review-rate assumptions will be tested, and discounted. The Court rejected estimates of 25 documents per hour as too conservative, stating that experienced reviewers should average at least 30. Generic industry metrics, unsupported by case-specific evidence, are increasingly vulnerable in both regulatory and civil contexts.
Seventh, reuse of prior review work should be expected where feasible. The Court placed weight on the overlap with related U.S. FTC litigation and questioned why earlier collections and deduplication were not leveraged more aggressively. In complex civil litigation, the same expectation will arise wherever parallel proceedings or prior matters exist. For institutions involved in multiple actions, this brings into play policies and best practices for data management, preservation and retention.
Eighth, parties cannot impose their own internal timelines on the adjudicative process. One of the sharpest criticisms was that Amazon appeared to staff and budget the review to meet a preferred eight-month schedule, rather than scaling resources to meet the Court’s order. That lesson applies squarely to civil litigants who attempt to manage discovery by cost-containment rather than compliance.
Ninth, team size matters, and courts will scrutinize it. The Court did not accept that modest team sizes were justified where millions of documents were at issue, particularly for a well-resourced party like Amazon Canada. Under-resourcing is increasingly framed as a choice. In other words, when courts assess reasonableness, budget/resource constraints may be dismissed as excuses.
Tenth, delay itself can constitute prejudice. In the competition-law context, the Court emphasized harm to the public interest. In civil litigation, the same reasoning translates into prejudice to opposing parties and to the administration of justice. Delay is no longer neutral simply because liability has not yet been established.
Taken together, the decision underscores a broader trend. Whether the matter is regulatory or civil, courts expect anticipation, proportional staffing, early use of analytics, and disciplined execution. Large productions are now treated as operational exercises, not excuses for extended timelines. Litigators who plan accordingly will retain credibility. Those who do not should expect increasingly little sympathy from the courts.
