Kim v. Cushman & Wakefield: A Federal Court Confirms That Email Search Terms Don’t Work for Microsoft Teams

By John Patzakis

A recent decision out of the Central District of California should be required reading for any legal team that includes Microsoft Teams as data source in their discovery plan. In Kim v. Cushman & Wakefield U.S., Inc., 2026 WL 1353455 (C.D. Cal. Apr. 24, 2026), the court held that search terms that may be appropriate for email may not be sufficient for shorter, less formal communications on a collaboration platform like Teams.

The plaintiff, Ms. Kim, alleged pregnancy discrimination after being terminated upon her return from maternity leave. The defendant asserted the termination was part of a reduction in force; Ms. Kim alleged that rationale was pretextual. The discovery dispute arose when it emerged that the defendant had not searched Microsoft Teams at all—even though, as one of the defendant’s own witnesses testified, Teams was one of the primary communication methods used at the company. To its credit, upon discovering the gap, defense counsel immediately ran the existing email search terms against Teams and produced 47 pages of messages, two of which proved relevant to the pretext analysis.

That partial cure satisfied no one. The plaintiff demanded a nearly indiscriminate search of “all reasonably likely repositories,” while the defendant maintained it had already run the terms against Teams and “there’s nothing left.” The court’s response: “Neither position is quite right.”

The Teams Ruling: Keyword Searches Alone Are Not Enough

The heart of the opinion is the court’s recognition that rerunning email-oriented search terms against Teams data is structurally flawed. The defendant’s terms all required “Connie Kim” as an anchor—e.g., “Connie Kim” NEAR “terminat!”. As the court explained:

“It is arguable whether that may work well enough even for emails, but it cannot work for MS Teams chats about transition planning among managers who might say ‘the Smartsheet’ or ‘Brooke’s workload’ without mentioning Plaintiff by name. Keyword searches alone, without more advanced and thoughtful search techniques, will be inadequate for Teams data—a medium where conversations are shorter, more informal, and less likely to include full names than email.”

The court also underscored the certification obligation that attaches once a party elects to search: “An objecting party that elects to search and produce—rather than move for a protective order—undertakes an obligation to search reasonably. See Fed. R. Civ. P. 26(g)(1)(B).” And the Rule 26(b)(1) proportionality analysis weighed in the plaintiff’s favor as to Teams, since the messages already produced confirmed that relevant communications existed in that repository.

Notably, the court declined to dictate methodology, holding that how the defendant fulfills its supplemental search obligation— “whether through custodian-based collection, refined keyword queries, or technology-assisted review—is Defendant’s choice, so long as the search is reasonable and the production is complete.” The court also traced the root cause to a pro forma Rule 26(f) conference: had the parties conducted a substantive ESI conference identifying repositories, custodians, and communication platforms at the outset, the Teams gap would have been caught months earlier.

In his excellent writeup of this case, Michael Berman of E-Discovery LLC consulted eDiscovery expert Tom O’Connor of the Gulf Coast Legal Technology Center, who raised a critical practical question: what tool was actually used for the search? O’Connor explained that while keyword searches inside Teams work, Teams supports only basic keyword matching and a few command-style filters. Per O’Connor, the native “Teams search indexes chat differently than email,” in that it:

• “Prioritizes exact word matches;
• Does not index message metadata as richly as Outlook;
• Often misses partial-word matches; and
• Returns fewer results when the term is too specific.”

In other words, even well-crafted Boolean terms can silently underperform when run against Microsoft’s native Teams index.

Why Kim Illustrates the Case for X1 Enterprise

The Kim decision validates what we have long argued at X1: when addressing MS 365 data for eDiscovery, the search methodology applied to it must be purpose-built. As we detailed when we launched our advanced MS Teams support, X1 Enterprise enables a targeted, iterative search and collection of Teams data in-place, with the ability to target individual custodians and specific messaging threads—displacing any need to mass download channels—plus unified search across Teams, OneDrive, SharePoint, Mail, laptops, and file shares, and one-click upload into Relativity for review.

Critically, X1 does not rely on the limited native Microsoft Teams index that O’Connor describes. X1’s patented technology builds its own full-featured index of Teams data, enabling precisely the “more advanced and thoughtful search techniques” the Kim court demanded. That includes detailed Boolean queries with nested operators, proximity, and wildcard/stemming support that execute consistently across both email and chat data—so counsel is not forced to choose between Outlook precision and Teams looseness. X1 also includes the ability to search on emojis, which is critical for Teams and other chat platforms, where a reaction emoji may be the entire substance of a manager’s response to a message about a “transition plan.”

X1’s patented in-place search and classification capabilities extend this further. Through the X1 API, organizations can programmatically execute searches and apply AI-driven classification models directly where the data lives—before anything is collected. Applied to the Kim fact pattern, that means counsel can iteratively test and refine looser, Teams-appropriate search terms against live data, measure the results, and classify what comes back—building a defensible, documented search methodology of exactly the kind the court invited when it referenced “refined keyword queries” and “technology-assisted review.” And because it all happens in place, the proportionality benefits are built in as only potentially responsive data is collected.

The lesson of Kim is straightforward. Courts now expect parties to identify collaboration platforms like Teams at the Rule 26(f) stage, to search them with techniques suited to informal chat data, and to do so reasonably and completely. Meeting that expectation requires solutions designed for the job.