## Facts

- A contractor and a surety litigated a multi-million-dollar dispute arising from alleged defects and delays on a major public construction project in the Bronx.
- Significant project records were maintained as electronically stored information (ESI), much of it held by the project construction manager, a non-party with extensive databases and email.
- During discovery, the parties agreed to use keyword searches to identify responsive ESI but could not agree on appropriate search terms or how to structure the searches.
- One proposed search set was narrowly drawn and likely to miss responsive documents; opposing proposals were broadly drawn and likely to generate large volumes of irrelevant material and unnecessary review costs.
- Counsel had not sufficiently consulted the data custodians and information-technology personnel to learn the terminology, abbreviations, and document practices used in the project records.
- The dispute was presented to a magistrate judge for resolution as an e-discovery methodology issue.

## Issues

1. What process must parties use under the Federal Rules of Civil Procedure to select, test, and refine keyword search terms for ESI production?
2. When may a court reject proposed keyword searches as unreasonable because they are likely to be under-inclusive or over-inclusive?

## Decision

- The court declined to adopt either side’s proposed keyword searches because they were not shown to be reasonable and risked significant under-inclusion or over-inclusion.
- The court directed the parties to follow a structured, cooperative process to develop search terms.
- The court required the parties to consult with knowledgeable custodians and IT personnel to identify how relevant concepts and project matters were actually described in the ESI.
- The court required testing, sampling, and iterative refinement of proposed terms to improve accuracy and control cost and burden.
- The court treated the dispute as one of discovery reasonableness and proportionality and intervened to require a defensible search methodology.

## Legal Principles

- Keyword searching is not self-validating; search quality depends on careful term selection, testing, and quality control.
- Discovery search protocols that are clearly under-inclusive or overly burdensome may be rejected as inconsistent with reasonable and proportional discovery obligations.
- Counsel must make informed decisions about ESI search methods by consulting custodians and IT personnel and learning the client’s terminology and data systems.
- Cooperation between parties is expected in developing ESI search terms, including sharing proposals and refining them based on testing results.
- Sampling and iterative adjustment of search terms are appropriate tools to reduce false positives and false negatives and to support a reasonable discovery process.

## Conclusion

The court resolved an ESI search-term impasse by rejecting untested, poorly designed keyword proposals and ordering a cooperative, empirically tested, iterative approach grounded in custodian and IT input and guided by proportionality and reasonableness in discovery.


William A. Gross Constr. Assocs., Inc. v. Am. Mfrs. Mut. Ins. Co., 256 F.R.D. 134 (S.D.N.Y. 2009)