TL;DR
In EEOC v. GEM Management, LLC, the Eastern District of Kentucky criticized a discovery process that treated ESI collection as a casual, unsupervised exercise, citing unreasonable search methods, apparent self-collection, and a defective production format. The ruling is a reminder that letting custodians gather their own data without a documented, supervised process tends to erase whatever it saved the moment a court starts asking questions.
What Happened in EEOC v. GEM Management?
The Eastern District of Kentucky granted the EEOC’s motion to compel on May 15, 2026, after finding that GEM Management’s discovery process failed at nearly every stage. The court’s language was direct. It described GEM’s approach to searching email and computer systems as little more than a casual “Ctrl+F” exercise among current employees, and found the company’s search methodology, undisclosed search terms, and undisclosed custodians manifestly unreasonable. When GEM argued that a third production attempt would be costly, the court’s response was equally direct: the company should have done it right the first time.
What Search Failures Did the Court Identify?
The court found that GEM used employees’ names as search terms rather than properly identifying them as custodians, and that the company appeared to engage in self-collection without supervision. The search also omitted terms an adequate inquiry would have caught, including the alleged harasser’s own nickname. Compounding the problem, GEM’s written explanation of its methodology relied on the passive voice throughout, describing what “was requested” rather than stating who searched what, with which terms, and over which date range. The court read that vagueness for what it was: an inability to account for the actual scope of the search.
Why Did the Production Format Fail Too?
The production failed because GEM delivered PDF files without the DAT load files that the court had already ordered. The EEOC had specified single-page TIFF images with a delimited DAT file containing designated metadata fields, consistent with its rights under Rule 34. GEM’s PDF-only production did not meet that standard, and the court rejected GEM’s attempt to shift the cost of a corrected production back onto the EEOC. In this case, search failures and production failures were not separate problems. An unsupervised, undocumented collection process rarely produces a clean, properly formatted output at the other end. One failure created conditions for another.
Why Does Self-Collection Keep Producing This Outcome?
Self-collection removes the documentation and oversight a defensible process depends on, and courts have been saying so consistently for years. GEM Management did not break new legal ground. It reaffirmed a standard the Southern District of Florida set out in EEOC v. M1 5100 Corp., 2020 WL 3581372 (S.D. Fla. July 2, 2020), where the court held that attorneys have a duty to oversee a client’s collection of ESI because self-collection by a layperson raises a real risk that data could be destroyed or corrupted in the process. That duty runs through Rule 26(g), which requires an attorney to certify, after reasonable inquiry, that a party’s discovery responses are complete and correct. An attorney cannot honestly make that certification without knowing how the underlying collection was conducted.
Many argue that the problem is limited to what a custodian collects; removing the air of selectivity and potential obfuscation still does not diminish the range of problems that can occur. Outside of missing documents and data to collect, the problem also extends to what the process fails to document. A defensible collection must identify the custodians, data sources, date ranges, search terms, tools, export methods, and any errors or exceptions encountered. In GEM Management, the court noted that the producing party had not clearly disclosed its custodians, date limitations, or search methodology, leaving counsel unable to explain how the collection had been performed or why known responsive documents were absent. Without a reliable audit trail, the adequacy of the process cannot be meaningfully tested or reproduced.
Self-collection, for many custodians, also tends to focus on the records a custodian can readily identify and locate rather than the systems in which responsive information may reside. Relevant ESI may be distributed across email accounts, local devices, shared drives, cloud repositories, messaging platforms, mobile devices, and archived or deleted locations. In GEM Management, the process appeared to rely in part on searching employee names as terms rather than treating those individuals as custodians whose actual data sources required collection. That distinction is most significant. A keyword search may identify references to a person, but it does not replace the systematic identification and collection of that person’s email, files, devices, and other potentially relevant sources.
There is also a material difference between preserving visible or easily accessible content and preserving the underlying electronically stored information. An email includes more than the text displayed on the screen, and a spreadsheet includes more than its printed appearance. Message identifiers, date and time stamps, attachment relationships, formulas, text in graphical form, hidden content, file paths, and other metadata may be necessary to authenticate, organize, search, deduplicate, and understand the significance and validity of these records. The production issues in GEM Management, including emails produced as PDFs, the absence of a usable load file, missing previously produced pages, and inaccurate or nonsequential Bates numbering, illustrate how deficiencies at collection and processing can carry directly into the final production.
Self-collection introduces judgment calls that a court can never fully verify after the fact. A custodian deciding what counts as relevant is making a legal determination, not a clerical one. Search terms applied inconsistently across custodians tend to yield inconsistent results. Such inconsistency alone can draw judicial criticism, even without any finding of bad faith. Metadata is especially exposed during self-collection, since forwarding an email, exporting a file via a native sharing tool, or printing a document to PDF can permanently strip metadata, often without anyone noticing until review is already underway. Production problems tend to trace back to the same root cause. When collection is not standardized, the resulting production is rarely standardized either. This is exactly the pattern GEM Management shows from searches through Bates numbering.
What Does a Defensible Collection Process Look Like?
A defensible collection process applies a single documented method consistently across all custodians and preserves native data and metadata from the moment of collection. Search decisions get recorded as they are made, not reconstructed later from memory. Chain of custody is built in real time rather than assembled after a motion to compel forces the issue. That kind of process is what turns a claim of “we collected the right files” into something a court can actually verify, rather than something opposing counsel can dismantle in a hearing.
The financial case for this is not abstract. GEM Management now faces a second collection effort, a corrected production under court order, and legal fees for a dispute that a documented process would likely have prevented. Whatever a company believes it saves by skipping proper collection tends to disappear the moment a court orders a redo, and by then the fees attached to fixing the problem generally outweigh whatever the original shortcut was meant to save.
How Can Complete Legal Help with Defensible Collection?
Complete Legal’s forensic collections team builds a documented, consistent process across every custodian from the first device or data source touched, preserving native data and metadata and creating a chain-of-custody record the moment collection occurs. That is the foundation on which search, review, and production all depend, and it is the piece GEM Management shows is going wrong from the start. If your team currently relies on custodians to collect their own data, or you are not certain how your process would hold up under the kind of scrutiny GEM Management received, reach out to Complete Legal to talk through a defensible alternative.
Contact Us:
📧 info@completelegal.us | 📞 888-885-9909
Key Takeaways
- EEOC v. GEM Management, LLC, 2026 WL 1363162 (E.D. Ky. May 15, 2026), is a process-failure case. The court found deficiencies in search methodology, custodial identification, apparent self-collection, and production format in a single ruling.
- The court’s own language, describing the search as a casual “Ctrl+F” exercise and stating that GEM should have done it right the first time, sets the standard courts expect going forward.
- Metadata is especially vulnerable during self-collection, where forwarding, exporting, or converting a file to PDF can strip it without anyone realizing it until later.
- Search failures and production failures are connected. An undocumented collection process rarely produces a properly formatted output at the other end.
- Redoing the collection after a motion to compel typically costs more in fees and delay than a documented process would have cost from the start.
Related Articles:
- Best Practices for ESI Collection and Modern eDiscovery Challenges
- Why Hiring a CCE-Certified Computer Examiner Matters in Digital Forensics and eDiscovery
- How to Avoid FRCP 37(e) Sanctions for Metadata Spoliation
