Privilege Logs:  New Techniques to Achieve Proportionality – The “Certification Log”

Maryland’s Proposed Data Privacy Act – Part of a Trend?
March 2, 2024
Failure to Show “Intent to Deprive” Leads to Denial of Rule 37(e)(2) Sanctions
March 5, 2024


Privilege logs have been getting a lot of attention, with good reason. “Privilege logging is arguably the most burdensome and time consuming task a litigant faces during the document production process.”  The Sedona Conference, “Commentary on Protection of Privileged ESI,” 17 Sed. Conf. J. 97 (2016).  Likely that is due to the complexity and fallibility of search, as well as the potentially serious consequences that may follow production of privileged ESI.

In Privilege Waived on WhatsApp Audio Files Due to Lack of Privilege Log ( (Feb. 27, 2024), Doug Austin described Coker v. Goldberg & Assocs. PC, No. 21-CV-1803 (JLR) (BCM) (S.D.N.Y. Jan. 24, 2024), where the court found privilege waived by the failure to provide a log.  Accord, Kelly Twigger, Can Failing to Produce a Privilege Log Lead to Waiver of Privilege and Sanctions? | Association of Certified E-Discovery Specialists (ACEDS) – JDSupra (Feb. 15, 2024)(“The Court said you haven’t produced a privilege log, therefore, you have waived the privilege as it exists in these communications.”); J. Leonard & T. O’Connell, Chancery Finds Wholly Generic Objections to Discovery Requests Result in Waiver and Fee-Shifting | Morris James LLP – JDSupra (Feb. 23, 2024).  In Mancia v. Mayflower Textile Servs. Co., 253 F.R.D. 354, 359 (D. Md. 2008), general “to the extent” objections were called “worthless” and could be viewed as a waiver.

Rules that require description of information withheld under a claim of privilege frequently do not specify what a log must contain.  E.g., Fed.R.Civ.P. 26(b)(5)(A); Md. Rule 2-402(e)(1).  They are often supplemented by guidelines and pattern discovery. E.g., D. Md. Discovery Guideline 10; Md. Discovery Guideline 5.

Former President Reagan said that negotiations are a process of “trust but verify. It’s still play, but cut the cards. It’s still watch closely. And don’t be afraid to see what you see.”

That is also the role of privilege logs – – “trust but verify,” and “play, but cut the cards.”

While clawback agreements[1] and non-waiver orders[2] may remedy imperfection after-the-fact, timely and detailed objections accompanied by a timely and specific privilege log are the first step in preservation of privileges.

Fed.R.Civ.P. 29 and Md. Rule 2-401(g) permit stipulations governing discovery procedures.  Two recent, comprehensive publications provide some excellent suggestions and samples of privilege logs.  The Sedona Conference, Commentary on Privilege Logs (Feb. 2024, Public Comment Version)(citing Rule 29);  EDRM Privilege Log Protocol: Version 2.0 (2023).  EDRM has also posted sample logs and exhibits at Privilege Logs – EDRM.  Sedona has posted samples at Exemplar-privilege-logs.xlsx (

Generally speaking, there are four types of privilege logs.

First, there is the traditional log, sometimes called a “Vaughn Index.”  It is a document-by-document listing of materials withheld under claim of privilege.  While it has a long history of acceptance, everyone recognizes that it may be costly to prepare.  Proportionality is always a factor to consider.  Further, there are many other problems with document-by-document logs. For example, in Hon. John M. Facciola & Jonathan M. Redgrave, “The Facciola-Redgrave Framework,” 4 Fed. Cts. L. Rev. 20 n. 55 (2009), the authors conducted an informal survey of Judges and attorneys.  The result showed widespread disagreement over the amount of detail required in such a log.  Sedona describes the “descriptive narrative” as “one of the most contentious aspects of privilege logs.”

Second, given the use of litigation review databases, “metadata logs” have, in my experience, achieved acceptance.[3]  I have used them for years.  Essentially, the review database is set up by the user to display fields that parallel the traditional log, such as “date,” “from,” “to,” “cc,” “bcc,” “subject,” “custodian,” “author,” etc. Those fields are populated by metadata obtained from native files or load files.  A field indicating the type of privilege, e.g., attorney-client, spousal, accountant, joint defense, or work product, is added and populated during the review process. The user then exports those fields to a spreadsheet and, voila, a metadata log has been created.  That is much less expensive than a traditional log; however, it may also have shortcomings.  To give only one example, the “subject” of an email may not comport with its contents, may not provide sufficient information, or may provide too much information by disclosing privileged information.  Sedona and EDRM suggest agreement to a sampling process in which the log recipient has a recognized right to seek additional information for X% of the log entries.

Third, a “metadata plus log” is a variant that adds a “field” that may be manually populated during the review process.  Both EDRM and Sedona describe it as the preferred type of log.  During review, the reviewer types an explanation of the reason each document is withheld.  For example: “Memo made at direction of counsel and sent to counsel for purpose of seeking legal advice regarding medical procedure.”  That description was deemed sufficient in Spilker v. Medtronic, Inc., 2015 WL 1643258, at *6 (E.D.N.C. Apr. 13, 2015).  Of course, while the manual entry of the “plus” field provides additional verification, it also adds cost.[4]

Fourth, a cost-saving feature is a “categorical log.”  These logs are authorized by procedural rules.  See Categorical Privilege Logs: Don’t Shoot a Mouse with an Elephant Gun.  For example, a typical agreement is that communications with outside counsel after the dispute became anticipated, or after suit was filed, need not be logged.  Another example is that communications with counsel in other litigation, or in other transactions, need not be logged.  Usually, the number of documents withheld must be disclosed and sometimes there must also be a certification that the privilege was not waived by disclosure.  However, defining categories may be difficult.  One interesting category suggested by EDRM is: “For corporations that have different matters involving same custodians over time, use of hash value to identify privileged files from other matters….”  While these logs are useful, in my experience, only a relatively small number of documents are covered by the categories.  Further, there is a lot more “trust,” than “verification.”[5]

Fifth, the Sedona “Commentary on Privilege Logs” describes a more novel type of log, called a “certification log.” Sedona wrote: “A privilege log is not the only option… for expressly making a privilege claim.” Id. at 9 and n. 26.  Sedona cited “Genesco, Inc. v. Visa U.S.A., Inc., 296 F.R.D. 559, 582 (M.D. Tenn. 2014) (plaintiff’s counsel submitted affidavits and other documents in lieu of log);[6] Fifty-Six Hope Road Music, Ltd. v. Mayah Collections, Inc., No. 2:05-cv-01059-KJD-GWF, 2007 WL 1726558, at *6-8 (D. Nev. June 11, 2007);  Hon. John M. Facciola & Jonathan M. Redgrave, The Facciola-Redgrave Framework, 4 Fed Cts. L. Rev. 20, 47 (2009) (advocating for production of an affidavit by the responding party that attests “to the facts that support the privileged or protected status of document and ESI within that category”).

It is instructive to consider the precise requirements imposed in Fifty-Six Hope Road, 2007 WL 1726558 at *8:

Accordingly, the Court will not require Plaintiffs to produce a privilege log for each allegedly privileged email communication, subject to Plaintiffs’ counsel submitting an affidavit or declaration under oath which states and describes the following: (1) That Plaintiffs have made a diligent and good faith effort to locate and produce all relevant and non-privileged documents, including emails, responsive to Defendant’s requests. (2) The number, or a reasonable estimate of the number, of the privileged email communications that exist. (3) That Plaintiffs have reviewed the alleged attorney-client or work-product privileged emails to ensure that relevant, non-privileged email communications are not being withheld from production and that Plaintiffs’ counsel verifies that no arguably non-privileged email communications are being withheld. (4) In the case of emails as to which the attorney-client privilege is claimed, the affidavit or declaration should include a verification that the emails were not provided to persons other than the client and attorney. If such communications were provided to non-clients, and the attorney-client privileged is still claimed, then a privilege log consistent with Diamond State for each such communication should be provided. (5) In the case of attorney-work product, the privilege may extend to persons other than the attorneys or the client, such as investigators. To the extent any attorney work-product emails have been provided to persons other than the attorneys or the client, an appropriate privilege log consistent with Diamond State should be produced for each such communication and an explanation provided as to why the work-product privilege applies.

I suggest that a modification of this “certification” approach may provide both “trust” and “verification,” while achieving cost savings.

I start by acknowledging that the modification: (1) relies on “search” and review; (2) no search and no review is perfect; and, (3) sometimes there are software glitches that result in errors.[7]  However, proportionality is as or more important as “perfection.”[8]

Further, under Rule 26(f)(3)(B), a discovery plan may provide for phased or focused and limited discovery and this phased approach has characteristics similar to those proposed by both EDRM and Sedona.

Consider a hypothetical in which a producing litigant approaches the opposing, receiving counsel and suggests that Fed.R.Civ.P. 29 permits the parties to stipulate to the following privilege assertion process:

  • In Phase 1, producing Counsel recommends an initial agreement on a Rule 26(b)(5)(B) and Fed.R.Evid. 502 order as a backstop.
  • The date range may be narrowed to a date (1) when litigation became reasonably anticipated, depending on the facts, and (2) end when litigation was commenced, because communications after the ending date could be categorically logged.
  • Counsel transparently discloses the “richness” or “prevalence” of privileged documents found in the database at 8%, using a simple random sample at 95% certainty with a 2% margin of error based on “eyes on” review of the random sample.
  • Counsel discloses privileged email names and domains and negotiates any requested modifications.
  • Counsel produces a “hit” or search term report, and states that when the email names and domains were run on the document universe, the unique “hit rate” closely approximated 8%, providing a degree of validation of the search.
  • Counsel discloses the number of unique documents that will be withheld.
  • Counsel suggests that all hits be considered presumptively privileged and a stipulated, but simpler, metadata log would be provided, with the stipulation to be submitted in advance to the court for entry as an order.[9]
  • The agreement would provide for a Phase 2 “meet and confer” in which the recipient could question specific documents and the producing party provide explanations under the non-waiver protection of Fed.R.Evid. 502(a), governing intentional disclosures.[10]
  • The agreement could provide that the recipient could demand a metadata-plus log for up to X% of the withheld documents.

While admittedly imperfect, this process could provide satisfactory verification.  It would likely reduce costs.

The five types of privilege logs are not mutually exclusive.  A litigant could negotiate a log, for example, that is part categorical and part metadata plus.  Or, there could be a phased agreement to a metadata log combined with a duty to meet and confer about any questions and an obligation to provide a metadata plus log upon request for up to X% of the documents. And, as Sedona points out, there can be different types of logs for different types of ESI, such as Slack channels and texts. Sedona at 39.[11]

Both EDRM and Sedona stress the need for cooperation.[12] For example, EDRM suggests that, once parties begin review, “they shall promptly notify opposing parties of any unique or ‘gray area’ issues that could be resolved up front to reduce the likelihood of later disputes or having to re-do logs later.”  This would include, for example, whether in-house counsel was acting in a non-lawyer capacity and whether there are third persons to whom disclosures were made but who are not considered “privilege breakers.”  Id.

ESI comes in many “flavors,” Lorraine v. Markel Am. Ins. Co., 241 F.R.D. 534, 538 (D. Md. 2007), and one size will not fit all.  Sedona and EDRM both discuss the problems of logging email chains, inclusive emails, attachments, and other varied forms of ESI. Accord, Facciola, 4 Fed. Cts. L. Rev. at 38.

EDRM provides some excellent examples of “name normalization” issues at EDRM-PrivilegeLogProtocol-Background_ABC.pdf at 4; accord Sedona at 21.

The EDRM and Sedona papers provide an excellent description of the types of logs, the advantages and disadvantages of each, ways to address complex ESI such as email chains, inclusive emails, texts, and collaboration channels.

This blog was initially posted on  Electronic Discovery Reference Model and  JD Supra.


SUPPLEMENT: See Doug Austin, In Camera Review by Special Master Ordered by Court ( 26, 2024).

UPDATE:  See Doug Austin, Privilege Logging is Unwieldy Thanks to Technology. Can AI Help? ( 19, 2024)(“’When I got on the bench in 1995, the privilege logs in a typical case [were] two to three pages, maybe 50-100 entries,’ former U.S. Magistrate Judge Andrew J. Peck said during the Legalweek 2024 Judges Debate keynote in January. ‘Now the privilege logs are like little novels, and there may be 10,000 or more entries. That is very expensive and is often useless to the other side in figuring out what is or isn’t privileged.'”).

[1] Fed.R.Civ.P. 26(b)(5)(B); Md. Rule 2-402(e)(3).

[2] Fed.R.Evid. 502; Md. Rule 2-402(e)(4).

[3] Sedona states that precedent is “mixed.”  While Sedona points to search issues, they are beyond the scope of this blog.

[4] Both EDRM and Sedona make clear that a “metadata search” is likely to miss privileged emails that are forwarded to non-attorneys who are within the privilege.  Thus, a metadata log should not be populated by a search that is limited to metadata fields.

[5] Frequently, one category of documents excluded from the logging requirement is redacted documents, if two conditions are met.  First, the redaction box must state the privilege being asserted. That can be easily set up in the litigation review software.  Second, the litigation review software must provide a code or tag that enables the recipient to search for redacted documents.  That is provided by most commercial litigation review packages.

[6] That decision was superseded without relevant change in Genesco, Inc. v. Visa U.S.A., Inc., 302 F.R.D. 168, 191 (M.D. Tenn. 2014).

[7] What Does “The Making of a Surgeon” Have to Do With ESI and “Software Glitches?” (2011).

[8] Sedona yellow flags this conclusion at n. 17: “Practitioners should be aware, however, that the application of proportionality to privilege logs continues to be disparately examined by courts after undertaking varying levels of analysis.”

[9] While it may not be necessary, the order could be a “Hopson-immunized” order to further protect against waiver.  Under that process, if the court finds pre-production methods were reasonable, it can order production.  If a privileged item slips through, there is no waiver because the production was not voluntary, it was ordered. See M. Berman, “The Continuing Utility of the Hopson-Immunized Clawback,” in M. Berman, et al., eds., “Electronically Stored Information in Maryland Courts” (Md. State Bar Ass’n. 2020).

[10] For example, the EDRM Protocol suggests an agreement that: “Once any privilege log is produced, the requesting party shall within 30 days (or a reasonable period agreed upon by the parties), inform the producing party whether it would like to meet and confer to discuss the initial log. The requesting party may select a reasonable number of log entries to inquire about.”  The EDRM Protocol specifies the procedures to follow.  Note also that the Maryland Rules do not have an analog to Fed.R.Evid. 502(a).

[11] See “Modern Attachments” or “Pointers”- What is a Document? (Part IV).

[12] For more information, see Historical ESI Highlights – Part VIII – Cooperation and The Cooperation Proclamation.