Episode #6: How to Reduce eDiscovery Costs with Intelligent Archiving

Transcript:

Bill Tolson:

Welcome to the Archive 360 podcast titled Corporate Legal Budgets are Getting Squeezed, How to Reduce E-discovery Costs with Intelligent Archiving and Information Management. With me today again is Jim McCarthy. Jim is the chief compliance officer and general counsel for Archive 360. My name is Bill Tolson and I'm the VP of compliance and e-discovery at Archive 360. Welcome Jim.

Jim McCarthy:

Hi Bill, it's nice to spend some time with you today.

Bill Tolson:

Yeah. This is an interesting subject right down in your lane, as well as mine. You obviously, being a GC and me being involved with e-discovery for geez, 25 years or so. This is going to be fun. So let me kick it off. And as you can tell by the title, we're talking to you about corporate legal budgets, the cost of e-discovery going up. And one of the main reasons for that is corporate litigations being driven by the COVID-19 pandemic. And that's a big part of it, but there's also the ongoing litigation as well.

Bill Tolson:

And I think corporations, corporate legal folks, GCs are getting squeezed by the CEO and the board of directors about cost of e-discovery. And I'll get a little bit more into what the actual cost of e-discovery is, but it could be a very big part of a corporate legal budget. And it's hard to plan for, obviously, because you don't know when you're going to be sued. So we'll get into what general councils and corporate legal can do to number one, understand what's going on with e-discovery, e-discovery processing, and then what we can do to help at least get control of e-discovery budgeting, or even lower it in a lot of cases. So to better control corporate legal budgets and e-discovery response times, because it's not like you've got forever to respond to an e-discovery, many GCs focus on bringing more of the e-discovery process in-house to lower costs.

Bill Tolson:

And this has been going on for many, many, many years. I remember talking to you about it back in 2006, 2007, but we go through these cycles and I think we've gone through a cycle over the last several years where corporate legal, especially in small to medium-sized companies, but even in large, get overwhelmed and they don't necessarily have the personnel or expertise to respond. So they end up basically telling or asking their outside counsel to do the e-discovery for them, which law firms do that all the time, but you're paying law firm prices when you do that. And that'll be one of the topics we talked about further into the podcast, but in reality, there are things that corporate legal can do prior to moving stuff off to external counsel, to drastically affect the cost of e-discovery.

Bill Tolson:

So I get ahead of myself a little bit, but in this podcast, Jim and I are going to explore the issues around current corporate e-discovery practices, including the rising cost and the reliance, total reliance in a lot of cases, on external offerings for e-discovery response. Additionally, we're going to focus on how those costs can be brought better under control and like I said, even reduced by utilizing information management and archiving solutions to help companies do more of a proactive discovery and proactive culling before they bring their outside counsel into it. And Jim, as we get into this more, you better than I know the costs associated with external law firms and how they run discovery and the responsibilities that they have, that they have to answer to the judge too, in responding to e-discovery for one of their clients, correct?

Jim McCarthy:

Yes, Bill, some people refer to this as the, you mentioned the over collection and then not purchasing the technology as the two biggest mistakes that many a GCs could make. But what I would like to suggest is that not so much a mistake as a purposeful reticence to adopt the technology. And I think what we're experiencing now is a rather longer arc of early adopters to predictive coding or technology assisted review and discovery and I would suggest that that's purposeful. There's been a slow legal mainstream adoption of this, I think, for a very selfish reason, the truth is, is that outside counsel makes a tremendous amount of money in review of documents. The more documents you have, the more review, the more billable hours by younger associates, Bill. This technology has the effect of reducing those hours dramatically and therefore the costs to organizations. But you can see that outside counsel would have its business model upended if this technology assisted review or predictive coding was used in mass, right?

Bill Tolson:

No, that's a great point. And I've talked about this for years with various attorneys and obviously it's not something they want to publish, but I was involved in the early predictive coding days back in 2012, 2013, working for a company that was pioneering predictive coding. And by the way, for people who don't recognize that term, it's basically a machine learning based culling and review capability so that you train the computers to recognize potentially relevant documentation for the given case and then it calls down the result sets and actually does the reviews at a very high accuracy level. And that can use algorithms and all kinds of neat things to actually measure the accuracy.

Bill Tolson:

And we were normally at 98% accuracy rate on tagging responsive documents versus relying on humans that would do the review. I mean, you might have a team of 10 outside contract attorneys spread all over the world, working on a million document review case. And you're going to notice a consistency issue based on where those contract lawyers are located, what countries, what kind of school system, where they were involved in, the law school they graduated from, all the way down to how their personal lives were affecting their day to day. So we normally, and we measured this, we normally would see a 40 to 60% accuracy rate with human review versus going on the machines at a 98, 99% review. Right? So.

Jim McCarthy:

Bill, I could add some personal attestation to what you're saying is that, I was one of those army of young associates that were in a large litigation factory. And there's a lot of human element that goes involved. I mean, I can remember nights where we were up at three and four in the morning, going through documents tagging. You could imagine young attorneys, there's variables there, late nights, a lot of hours. And like you said, the direction that we were always given was when in doubt, or if you don't understand something, include it.

Bill Tolson:

Yeah. Which obviously raises the number of items to be reviewed. Back in the day, back in 2013, we were one of the pioneers in trying to get the US courts at least to start to adopt that predictive coding technology. And I'll tell you, it was a long slog. I mean the attorneys are not known as-

Jim McCarthy:

Early adopters.

Bill Tolson:

...early adopters and judges sure aren't either. There's a couple here and there that we could talk about, but you have to absolutely prove to them that this document sets review set that went through a predictive coding process is extremely consistent and highly accurate. And over time we started to get judges basically instructing the plaintiffs and defendants to use predictive coding to lower the overall cost. And like Jim said, using predictive coding, you could potentially lower the cost of an entire e-discovery process by 80%. And Gardner for example, several years ago, and some others, by the way, Cleaning Rand Corporation, estimated that the average cost of one e-discovery in the United States, and again, I stress average, was one and a half million dollars. So if he can knock 80% off of that, and the vast majority of that is reviewed by the way, that's sure going to help your legal budget exponentially.

Jim McCarthy:

It's been just about nine years now since I guess the first case where predictive coding was permitted and approved. And it's still, given how old it is, still not universally used Bill. And part of that, of course, the cost of implementing it.

Bill Tolson:

Like you say, attorneys don't necessarily want to give away a money-making process.

Jim McCarthy:

That's for certain.

Bill Tolson:

That surprises me a bit that it hasn't been more widely adopted because it is an absolutely proven technology. And like you said, I always refer to it as predictive coding because we coined the term, but technology assisted review, computer assisted review. There's all kinds of acronyms around it, but it is one of the tools that would help reduce the overall review set that would go directly to the legal departments, bottom line.

Jim McCarthy:

There's always another side to every position, right? And the defense counsel could certainly articulate to its clients that it would be concerned with a potential ruling by the court, either an adverse inference ruling of holding back material, discovery, or some sort of a lack of due diligence on the responding council's efforts as a mean to discourage the use of predictive coding or TAR. And I suspect that conversation happens frequently when clients are talking to their outside counsel about culling the sample set before they give it to them. So there is that legitimate concern that a human is not looking at it and we're relying on the technology too much. But over time, I think you mentioned it Bill, when that head to head competition, technology did not crumble. I mean, it was well-received.

Bill Tolson:

And again, mathematically it's absolutely provable, but the main things are it's because it's a computer programmer machine, it's constantly consistent. And it's provability of being wildly accurate is beyond argue now.

Jim McCarthy:

And Bill, it gets better, doesn't it? The same software when it's working on the set and the user starts to use it more often, the very nature of the artificial intelligence is that it gets better at identifying relevant documents, doesn't it?

Bill Tolson:

You're constantly going through training cycles and every training cycle basically makes it much more accurate.

Jim McCarthy:

And attorneys get better at it too. I mean, the more senior you get in your practice, you're actually able to go through documents and better. But unlike technology, when lawyers become more senior, they start charging more.

Bill Tolson:

That's actually an interesting case that we looked at back in 2013. You'd give a data set to your outside law firm, they'd give it to their lowest level associates or even contract reviewers to do, but then they'd charge you the partner price. So they might've paid $50 an hour to have associates or something do it, but their partner bills at $500 an hour. So they bill it at $500 an hour for the review.

Jim McCarthy:

The underbelly.

Bill Tolson:

Yeah. Well, and I sort of understand that, but by the way, while that was going on, there was a lot of costs pushback on law firms especially and law firms started to become pretty aware that legal departments were starting to shop around their legal needs just because of the ongoing costs. So law firms, many of them I actually did training with and stuff like that recognize that. And they would either give a single price for discovery, or they actually started using predictive coding and passing along a big, not all, but a big part of the cost savings to the client, just to retain them. They did recognize that clients were getting a little upset with the constantly rising legal costs.

Jim McCarthy:

It's harder to understand why more legal mainstream hasn't adopted this Bill. When you look at the e-discovery rules themselves, they underwent a massive restructuring in 2015, essentially. They had the effect of reversing a lot of earlier decisions that imposed very harsh sanctions on parties that did not properly produce discovery.

Bill Tolson:

One of the big ones in the 2015 amendments where it was the idea of proportionality.

Jim McCarthy:

Yeah. Round that out for those listening that may not.

Bill Tolson:

The original, like 2005, 2006 amendments to the federal rules of civil procedure that had pretty cut and dried black and white rules around what data needed to be collected, preserved, reviewed those kinds of things and there was a lot of questions around the inaccessibility of backup tapes and all the data on backup tapes and pre 2015, it was more of an open question and many smart plaintiff's attorneys could force the defendant to start restoring backup tapes to look for data. And in the 2015 [inaudible 00:14:18] CPM amendments, the idea of more of a proportionality question was allowed by the judge to say, do you have reason to believe that the defendant should go back and restore the 500 backup tapes? What data do you think was on there? And if you don't have a good reason for them to do that, because then the defendant usually is going to pay for the discovery of their content.

Bill Tolson:

So one of the tricks of the trade by the plaintiff is just to force the e-discovery costs to be as high as possible so the defendant settles, so they save money or something like that. And that was an ongoing strategy. But with the idea of proportionality, the judges has more leniency to say, "No, I don't believe that what you're going to get back from forcing the company to restore 500 backup tapes is going to mean much to the case. Therefore, I disallow the defendants having to restore backup tapes." Or any other kind of inaccessible data. But the 2015 amendments kind of centered in a little bit more around the anticipation questions, some of the things won't get into now, but the difference between the 2005 amendments versus the 2015 are really interesting.

Jim McCarthy:

Absolutely.

Bill Tolson:

All right. So Jim, you mentioned kind of the two potentially biggest mistakes that corporate GCs, corporate legal departments make when responding to discovery and you mentioned over collection. And over collection of data is basically two parts to this, one over collection, and then relying on their outside attorneys completely to do discovery. So the problem with over collection is the GC tells the IT folks, "Go collect all the data you can find on these 200 custodians, whatever it may be, give it to me." And then some companies, not all, but some companies will say, "Okay, I will turnover all of that to my outside counsel, let them worry about it." Well at the cost that we can get into a little bit later here, I mean, that's extremely expensive. If you turn over a million documents to your outside law firm, they're going to read each and every one of the one million documents to determine privilege or responsiveness or non-responsive.

Jim McCarthy:

And I will tell you Bill, having been on that team that would review that massive data set you get from your customer. Most of the documents, a large percentage are simply not responsive to that discovery request. And we would leave a lot of documents on the editing floor, whether it's by privilege or one of the other self-critical analysis, or simply not being relevant. And we would simply make a fortune on creating a privilege log and calling down that set to what's ultimately produced to plaintiff's counsel. So I guess what we're getting at here is that the clients now can do that initial culling by employing some of this relatively tried and true technology.

Bill Tolson:

Great point. Some large law firms would utilize some e-discovery platforms to cull the data down instead of having strictly manual on-track attorneys or in-house attorneys do it. I mean, for medium and small sized law firms, most of them didn't have, and probably still don't have e-discovery platforms to do it in an automated fashion. So as Jim is saying, and what we've mentioned before, that first mistake, handing over a million or 2 million, I've seen 5 million beta sets handed over to outside counsel for them to go through the review process. I'll jump ahead a little bit here. I've done a lot of research on this and what's the cost involved? Well say for a million documents, culling it down to the lowest and then doing the review, the average number of pages that a contract attorney can review in an hour for responsive [inaudible 00:18:23] privilege.

Bill Tolson:

That means they have to read it and determine is this responsive to the case, or is it something, is it confidential, whatever it happens to be, was anywhere from 45 to 55 pages per hour. And if you're paying one of the lower end men, thinking that I have a million documents, start doing the multiplication on that, even low cost contract review attorneys, you're looking at $50 per hour, in-house probably more. So you could see that the review cycle, and I refer back to the Rand report back in 2012 and they had fantastic report. And for those of you, I could send you the link, but they got very specific and they determined that the cost of e-discovery was divided up into three areas, collection, processing, and review. Review was 70% of the total cost. So the smaller amount of documents, either you're just using predictive coding or some of the other things that Jim has mentioned, using basic information governance techniques, you could lower that 1 million documents down to 300,000, then that's 700,000 documents that you review, you're not going to have to pay your review attorneys to actually read.

Bill Tolson:

And not even talking about the accuracy or consistency or anything like that. Just sheer cost. So 45 to 55 pages per hour, times 50, 60, $70 per hour for the actual review. Now you start to realize that that cost can be overpowering. I have one example. I have a e-discovery calculator that I put together and looking at various functions, you're looking at a cost before doing pre collection and culling of $2 million for a given case. And if the defendant had used the culling and pre-collection and stuff like that, or use predictive coding, you could be looking at anywhere from a total cost of 300,000 to maybe a million. So at the least a million dollar savings on a given standard case times the number of cases that a company deals with in a given year. Where I've worked with companies that receive four e-discovery requests per week. I don't know about you, Jim.

Jim McCarthy:

Yeah. Maybe not that frequent, but yeah, certainly public agencies that we worked for getting that kind of volume at least. But Bill one note, I noticed that you mentioned the 40 to 50 documents per hour, that your average associates can get through. That is true and that's when we used to measure these things as discrete documents. We printed them out and handed the young associate a pile and had them go through them, but a separate note on emails because it's not a discrete document, is it?

Jim McCarthy:

Let's say that I have a September 10th email in front of me and I have to review it. Well, that's one of your 40 to 50 documents, right? But think about it. You don't stop when that person signing that email signs off on the signature page. You have to look at the thread. And that thread means that I'm probably looking at documents over again. So I'm not reading 40 to 50 documents in an hour, I may be reading the same document 10 times if I'm looking at different threads. So that makes the slug even slower. So maybe I'm not at 40 to 50, maybe I'm at 10 to 20 now when you're talking about emails.

Bill Tolson:

And to even complicate that, you gave the example of an email, what if that email had an attachment?

Jim McCarthy:

That's it.

Bill Tolson:

What if that attachment was an Excel spreadsheet? And you're looking at single cell formulas. I mean...

Jim McCarthy:

And predictive coding can cut through that. It looks at not only the email, but the attachment.

Bill Tolson:

Also having the documents in an archive or information management system where all of that data is indexable, meaning you could do a search and it's going to search the attachments as well for any responsive content, like a keyword or fuzzy or whatever it happens to be, a document archiving system or predictive coding. But the archive can do that. You can create a result set, you do a search and it searches all metadata, the email body, all attachments, everything within the attachment. And it's going to give you a result set. That result, lets just say that's 450,000 documents. Then you could start doing the culling on it, the search within a search and using related search terms and stuff and basically bring down that 450,000 to maybe the 200,000 that are truly potentially responsive to the case. And that's what you give to your outside counsel. And by the way, and I know Jim, you would bring this up, all of that has to be audited and reported because your external counsel wants to feel safe in saying yes, they did it right to the judge.

Jim McCarthy:

Oftentimes they're certifying to the court that its client has done this according to Hoyle, according to the the best efforts of the discovery rules. So they have to be comfortable that primary data set has been culled in an inappropriate way.

Bill Tolson:

Doing that culling, obviously, you're going to potentially either manually or automatically tagging the emails and attachments as responsive, non-responsive, privileged or whatever tags you're going to use. And based on those taggings for example, you can immediately, automatically produce a privilege log, which some opposing counsel wants to see it when you hand over a discovery datasets.

Jim McCarthy:

I know you've done some work on examples of how you get the discovery requests and what the charges are at the end of the day.

Bill Tolson:

Yeah, yeah. I mean, it really gets down to, I mentioned that there were two mistakes. When we mentioned the over collection and the reliance on external counsel, the other one is not being aware of nor purchasing technology to help defensively reduce the size of the primary data set by doing culling and other stuff. And the culling, you're going to look for duplicates. You don't want two different attorneys reviewing the same document that came out of two different parts of the system times 10,000 or whatever. Also when you do a keyword search or something like that, you're going to find the computer files, logs, all of that kind of stuff that the system automatically generates and it might include some of the content that you're looking for, it's just going to show up. So again an attorney is going to be charging to actually look at that.

Bill Tolson:

So that second mistake is being aware of the technology that anybody can use to do this collection and pre culling to really target and reduce your overall e-discovery costs. And again, I went back to kind of the cost figures, the average cost of e-discovery is potentially in the one and a half million dollar range. If you're relying on manual review, cost can be pretty astronomical at 50, 60, $70 per hour and 50 or 60 pages per hour to review. And data sets usually, not all the time, but data sets for each discovery are usually large. For example generally speaking, the average amount of data collected per custodian, this is not everybody in the company by any means, e-discovery basically targets specific custodians usually. Maybe you have 10 custodians or targets. The average amount of data that IT is going to collect on him is around three gigabytes.

Bill Tolson:

How many documents are in a gigabyte? Well, boy, that goes all over the wall. I mean, I use conservative numbers anywhere from 3,500 documents per gigabyte to 12,000. So we're looking at three gigabytes and 3,500 documents, 10,500 documents per custodian. And if you have 10 or 100 custodians, as part of the discovery, you can see where these costs go. And it's amazing that just doing some pre discovery, basically collecting data as part of your normal business. And by collecting, I mean capturing data, putting it into an archive, for example, for ongoing data management, you're consolidating all of your data in a single repository so that when you respond to discovery, you're going to search one repository, one or two, but I've worked with customers as a consultant where they were looking through 25, 30, 40 different repositories looking for data.

Jim McCarthy:

And it would be very persuasive in court to be able to present to the court that your data is consolidated in that way, where you're only looking in one or two silos, Bill, as opposed to disparate all over the place. We often talk about the case. One of the parties was sanctioned because they found, what was it, a backup drive under someone's desk at some point in time?

Bill Tolson:

It was a wall street bank. It was several years ago, probably six, seven, eight years ago out of Florida, but it was a financial and they kept doing discovery and they didn't have a program that consolidated information. So they kept finding data in various systems over time and they kept going to the judge and certifying that, "We're completely done. We did it correct." And then two weeks later, they'd have to go back to the judge and say, "Oops, we found some more data." And that kept happening. And finally, the judge just had enough of it and said, "You guys don't know what you're doing." They kept finding data, drives, backup tapes and stuff under desks in remote offices, all this kinds of stuff. So the judge finally said, "You're done, you can't do this." Judge issued an adverse inference.

Jim McCarthy:

That sounds like Judge Scheindlin from New York.

Bill Tolson:

Yeah.

Jim McCarthy:

And I know she's just retired a few years ago, but if it's the same case, I remember she had no less than five decisions in the Zubulake case.

Bill Tolson:

By the way, an adverse inference is basically an instruction to the jury, telling them, before court even starts, basically telling them, you can infer that the defendant who wasn't able to find all the data doesn't want you to see the data they won't produce. So draw any conclusion you want, but they're hiding data. I've had many lawyers tell me that's basically the kiss of death. You get an adverse inference and you've lost the case. And the only thing in question now is how many zeros you're going to be writing on the check.

Jim McCarthy:

Jury doesn't trust you at that point in time, you are toast.

Bill Tolson:

Yeah, it's kind of hard to come back from that.

Jim McCarthy:

Yeah. You were just speaking and reminded me of whether or not, I remember Judge Scheindlin's, one of her rulings that sending out a litigation hold letter was in and of itself negligence, per se. And despite the fact that she was reversed on that narrow point, it got me thinking, would it be malpractice for an outside counsel not to have its client cull that primary data set for a self-interested motive? Given the technology is advanced now.

Bill Tolson:

That's an interesting question. I think that most judges would not have an issue with that. Most judges are very, very aware of the rising costs of e-discovery and what discovery costs can do to the defendant, for example.

Jim McCarthy:

Coercive effect that outside counsel can have on its client, if outside counsel is directing its client to do this, to comply with the rules, it would be the rare client indeed that would push back.

Bill Tolson:

Yeah. I think judges are open to good faith efforts of reducing costs while responding to e-discovery and obviously these things you want to document, you want to talk about processes and procedures and technologies used. I don't think a judge nowadays is going to have an issue with the defendant's counsel saying, "The company GC oversaw much of the collection and an early culling. My law firm basically took it from there. We believe that the e-discovery processing was within industry expectations." I think it would be hard for a plaintiff's attorney to argue with that and say, "Well, no, the lead attorney didn't do it all, so therefore there's an issue." So I think courts are starting to, have begun to adapt to this kind of thing.

Bill Tolson:

So this is interesting and making sure that the the communication between the GC and the outside counsel is ongoing, and not that it isn't, I think probably 99% of the time, they're probably excellent at it, but just ensuring that documenting the discovery process. Number one, having a documented discovery process before you do discovery is always a good thing that you always rely on, but also following it and documenting how you followed it is going to mean a lot to a judge. I've seen judges look at these kinds of documents and say, "You were doing the best you can. I'm not going to fault you for that." Even when things happen, even when data is inadvertently deleted or things like that, if you can show good faith effort, then usually the court is not going to come down on you with a ton of bricks.

Jim McCarthy:

Most of the sanctions that you see is not because of the attorneys or the defense attorney's efforts, it's because of the client themselves not making adequate efforts to produce requested discovery.

Bill Tolson:

Or not even being aware of what kind of data that they actually hold. I mean, I've seen that many, many times. So tell their GC or even outside counsel, these are the repositories. Just an example, very quickly I was doing consulting work for a power distribution company and we were trying to consolidate this stuff. And when I first walked in, I said, "Okay, how many repositories, how many file servers do you have?" They said, "Approximately 1000 throughout the whole company." And they were relying on those 1000 to do discovery on and respond to SEC inquiries and stuff like that. After a month I found out they had 5,000 repositories and 4,000 of them they hadn't even been looking at for discovery. So that put the fear you know what into the juice.

Jim McCarthy:

Yeah. If you were at a deposition and that has to get revealed because you're under oath, that would be in and of itself a fun motion for plaintiff's counsel to have made.

Bill Tolson:

And all those past cases. I mean, boy, we've used a lot of time here, but go ahead. Do a closing comment there, Jim.

Jim McCarthy:

So I think what we're saying is that this is a sort of a call to action for GCs. Take advantage of the technology, specifically the predictive coding and technology assisted review that is out there. It's not just a matter of cost savings. It's more efficient. It's in fact, the state-of-the-art, you save yourself some money and you'll be more efficient at it with less likelihood of sanctions by the court. So you have to ask yourself why wouldn't you do it?

Bill Tolson:

Yeah. I mean, it really comes down to money savings. And for those who don't get to the point of using predictive coding, using information management techniques and archiving programs, many of them have this capability built in, the ability to do collections, search, the ability to do search within a search, the ability to cull down all in a legally defensible model and also maintaining chain of custody is always the biggest thing.

Bill Tolson:

So utilize technology, but also number one, don't rely completely on your outside counsel because they'll be happy to do it and they'll do a great job, but it costs a great deal of money, but utilize technology that you should be using any way to reduce your overall cost. And we can talk to you about that and show you some capabilities within our own systems that go directly to this piece. So with that, I thank you, Jim. And if anyone has any questions on the topic of e-discovery cost-savings, please send an email, mentioning this podcast to info@archivethreesixty.com and we'll get back to you or contact you. You can always check out our various blogs on the topic at our Archive 360 blog site. I do have several on this topic, so it's just www.archivethreesixty.com/slash blog to check that out and with that, Jim, thank you very much. It's been enjoyable as usual and thanks to everybody else for listening.

Jim McCarthy:

Bill, thank you very much. Good spending time with you today.