Introduction to Discovery (6/7): Estimate Checking is the Key to Controlling Costs (Part 1)

Although it is common practice to obtain "competitive quotes" in all business situations, the truth is that the discovery industry does not often lead to competitive quotes.

The reason for this is that the format of quotes varies from vendor to vendor, and there is no way to understand how to look at quotes and compare them. It is not that there are few competitors. Rather, the discovery market continues to grow in proportion to the increasing volume of data held by companies, and new discovery support vendors and vendors are springing up one after another. This seems to be the primary reason why companies are unable to successfully control the cost of discovery.

Is the difference between the quoted and invoiced amount an everyday occurrence?

In general business, whether in the manufacturing or service industry, it is rare for the quoted price to differ significantly from the actual price. If the actual invoice amount is significantly higher than the estimate, the person in charge of placing the order will be inconvenienced, and the contracted company may be accused of poor planning at the estimate stage, so if the price is slightly over the estimate, it will be kept at the estimated amount. Even when it is anticipated that the cost will exceed the estimate due to unavoidable circumstances, it is normal for the party receiving the order to contact the party placing the order and ask for permission.

However, in the discovery industry, there are often cases where the quoted price and the actual invoice price differ significantly. This indicates that the quotation is only a "reference price. It is not uncommon to compare the prices of multiple vendors at the quotation stage and think you have chosen the cheapest vendor, only to find that the final bill is more expensive than the other vendors.

Furthermore, it is difficult for companies to even judge whether the amount charged is reasonable or not. The fact that discovery fees are charged through an intermediary law firm, rather than directly from the vendor, is another factor that makes it even more difficult to confirm the details.

Even if there is a question about the amount charged, there is no way to ask a vendor other than the outsourced vendor about the appropriateness of the amount after the discovery is completed, asking, "As a third party, is this amount really reasonable? On the other hand, it is also unrealistic to ask the legal department of another company that has experienced litigation, "How much did your company charge for discovery?" It is also unrealistic to ask the legal departments of other companies that have experienced litigation, "How much did your company spend on discovery? As a result, discovery costs often begin and end ambiguously.

The key to getting the right estimate

Once you receive a complaint from the opposing party and decide to respond to it, your attorney will probably request that you take discovery measures. Upon receiving the request, the company's legal department will request quotes from discovery support vendors, and will select a vendor based on the amount and content of the quotes, as well as the vendor's track record.

The vendor selection will be made by the attorneys themselves (or, more accurately, not by the attorneys themselves, but by an in-house department called Litigation Support at the law firm to which the attorneys belong). It is important to note, however, that it is very risky to leave the selection of the discovery vendor to a U.S. law firm or a U.S. local subsidiary.

The reason for this is that it is difficult for companies with no experience in discovery to understand the breakdown of quotations and companies selecting vendors, making it difficult for them to make a judgment. If the vendor's work is of high quality, that is fine, but if it is not, the result is a contract for a higher-than-market price and low-quality discovery, which can lead to an unfavorable outcome in litigation.

In order to prevent such mistakes from occurring, the following section introduces the "Failure-Proof Estimate" method.

Quality Discovery Vendors have detailed estimates.

Example 1 is an actual quote used by one comprehensive discovery support vendor. For legal professionals who have no experience with discovery, this may be the first time they see this information.

We will provide a detailed explanation later, so we recommend that you first look at it as if you were selecting your own discovery vendor, taking note of your own findings and questions. Please take note of your own interpretations and questions.

Afterwards, the commentary below will help you understand which errors you are likely to make.

Items such as "data collection," "data process/analysis," and "data hosting" indicate the various steps in the discovery process. If the cost is calculated for each process, it is easy to understand the cost percentage of each task in the total cost, even if it is only a rough estimate.

Note that the "review cost" is the most expensive of all the steps. If you have good intuition, you may be able to determine that reducing the cost of this review would make a big difference in the overall cost.

In the discovery process, documents are collected, (documents) are analyzed, (documents) are examined to see if they (documents) are admissible as evidence, and (documents) are submitted. In other words, most of the subject matter is "documentation. Therefore, estimates are always calculated on a "unit price basis".

However, when it comes to calculating per document per copy, it is difficult just to find out the exact number, so we actually count "per computer" or "per hard disk". Or, in some cases, the calculation is made "per 1 GB" because the capacity of each piece of data differs.

A vendor with experience in discovery will know how much work to do, depending on the size of the company. The "estimated volume" is then calculated. In other words, the total cost of a discovery estimate is calculated as "unit price x estimated volume.

Some vendors do not provide this estimated quantity, and often provide only a "unit price" estimate. When we ask them why the total cost is not calculated, they usually reply that they cannot calculate the total cost because they do not know how much data they will need for discovery.

If you take this answer at face value, you can understand that "I see, if there is less data, the total cost will be less, and if there is more data, the total cost will be more.

When this is the case, the first thing that the legal staff thinks of as a way to reduce costs is to "reduce the amount of data. However, if you try to reduce the amount of data carelessly and delete items that should be submitted as evidence, you may be punished as a cover-up of evidence, so you must be very careful in this area.

The answer "I don't know the expected amount of data" can also be taken as "I don't have much experience in discovery" or "I don't want to disclose it as an estimate" if you look at it from a distorted point of view. The more rough the estimate is, the more excuses can be made for any discrepancy between the estimate and the actual invoice amount.

You should be especially careful about items that are calculated on a per hour basis. For example, when estimating the amount of a review, a vendor who provides data such as "reviewing 0 documents per hour" is conscientious. However, if the data is not clear, it is possible to "deliberately spend more time and raise the price".

In the case of a vendor whose work quality is low or who does not support Japanese, garbled characters may occur before the review, and the work may have to be redone as far back as the data processing stage. Many vendors will proudly charge the client for redoing work that was caused by the vendor's own mistake.

That is why it is a prerequisite that quotations are calculated on the basis of "unit price x estimated volume. It is not difficult for a vendor with a proven track record in discovery to estimate the estimated volume based on the size of the company.

In the quotation mentioned at the beginning of this section, "Note" includes a list of separately incurred costs. This is also a natural part of a general estimate, but some vendors may not include it at all.

For example, we have often heard stories of vendors adding one option after another that should have been explained in advance, such as "Documents created in Japanese will be recalculated based on a separate estimate," or "Analysis of e-mails sent and received via a mailer will be charged separately," resulting in a price that is far off the initial estimate. As a result, we often hear stories that the cost of the service is far off the initial estimate.

In many cases, the quotation does not include the costs of hosting, load file creation, and uploading, which are discussed later in this section. In addition, additional work to handle special file formats is often not included in the cost, so companies that handle such files are advised to check carefully.

How to View Discovery Estimates by Work Process

We have briefly explained the basic items and prerequisites for looking at estimates, especially how to compare them with those of other companies when getting quotes from them. From here, we will explain the details along with the example quotation we have just given you.

(1) Data Collection

Data collection, in the narrow sense of the term, is the process of duplicating and collecting data that may serve as evidence. In this estimation example, the process from preparation to replication and collection is summarized in a nutshell. The preparatory process is the identification and preservation of the data.

• The first step in identifying and preserving the dateline is to conduct interviews. Starting with the identification of the departments and parties involved in the lawsuit, information system personnel must be asked not only about the system and file formats, but also about data storage rules and regulations.
• The procedures for preserving data to prevent it from being overwritten, lost, or altered must also be discussed and agreed upon by the parties involved.
• It is also necessary to create a "data map" so that it can be easily identified during later discovery operations which persons have what materials and what data is stored on what media.

To perform this data collection, the vendor must send an engineer to the company. The work often takes several days, with additional travel and lodging costs. In addition, the "estimated volume" in the estimate is the cost of the work for the target device, and hard disks to store the duplicated data and to back it up may be billed separately.

(2) Data Process

Data process is a "data processing" operation that extracts necessary information from electronic data and converts the collected data into a database for analysis.

The amount of data preserved and collected in the previous process is enormous, and it is impossible for plaintiffs and defendants to view all of this data. Naturally, there is also a lot of data that is not relevant to the lawsuit, and such unnecessary documents must be sifted out quickly.

The data processing work includes seven types of data culling, filtering, and related tasks (e.g., preparatory work).

The following tasks (1) to (6) are preparatory tasks for (7) and the data analysis that follows. Culling is the process of removing completely irrelevant data in advance, and the most obvious example is the removal of program configuration files and OS configuration files.

Since the contents of compressed files cannot be checked as they are, they are extracted so that they can be checked, and if the same file is held by multiple parties or stored on multiple storage media, they are also excluded. After narrowing down the search to some extent, text information and metadata information are extracted, and then organized and stored in each created index.

Currently, most of the information disclosed in discovery is electronic data. If data stored on multiple computers, servers, and mobile terminals, and even archived e-mail data are included, the volume of information is enormous. As mentioned above, a printout of data from one computer would be equivalent to about four 2-ton truckloads of documents, so if there are 20 custodians (data holders), the documents would purely cover 80 truckloads of documents.

It would be virtually impossible for an attorney or staff lawyer to review all of those documents and identify whether or not to adopt them as evidence. Even if it were possible, it would require a huge amount of money just to pay the lawyers, plus a mind-boggling amount of time.

In addition, disclosing data that is not relevant to the case would clearly be detrimental to the company's strategy. Information about a new product should not be disclosed before it is launched.

That is why advanced IT technology is used to sift out documents that are clearly unrelated to the lawsuit. This sorting process is the "data process," and it is extremely important because the accuracy of this process will affect the accuracy of the subsequent discovery process.

(3) Data Analysis

Data analysis involves analyzing the data prepared in the previous steps of the process and preparing the data for evidentiary review. This is because even with detailed data process work, a large percentage of the data is still irrelevant to the litigation. Therefore, companies and their attorneys will select keywords and discovery support vendors will work together to provide technical advice.

Then, more advanced keyword searches are conducted to identify and extract the target data, narrowing the search to only the necessary data. This process is called "analysis.

There are three major types of work that occur in this process.

If the analysis is done properly and the necessary data is narrowed down, the next step, review, will be more accurate and efficient, and discovery itself should proceed smoothly.

What must be noted here is whether the vendor is processing for Asian languages as described in (2).

If the vendor does not support Japanese, problems may occur during the review process, such as "documents are garbled and cannot be viewed" or "documents are not sufficiently narrowed down," which may result in problems such as the review not being completed as planned.

There is a difference between an environment that "can read Japanese" and an environment that
"can handle Japanese.

It is also important to be able to support Japan's unique application culture. For example, it is believed that most companies use Outlook by Microsoft for mail software (mailer), but many companies in Japan use "Becky!" by RimArts or "Shuriken" by JustSYSTEMS.

Email information, as an external communication tool, is extremely important in discovery, yet if it is not compatible with these uniquely Japanese mailers, the accuracy of the process work is dramatically reduced.

However, it is better to check whether they can handle Japanese-language files in the first place, and if so, whether it is within the scope of the quotation or not, or whether they can handle Japanese-language applications as well, If so, is it within the scope of the quotation?

(4) Data Hosting

Data hosting is the management of data used in a review on a designated server so that it can be viewed by the review tool.

In the past, we have seen cases where the vendor with the lowest unit price at the quotation stage paid approximately three times the total hosting cost of the other vendor. The price varies greatly depending on what data is hosted and at what stage of the process.

Unfortunately, the discovery industry does not yet have a global standard for managing the discovery process. The idea that it is enough to disclose evidence as a result, no matter what stage of the discovery process the data is hosted at, is still prevailing in the industry. As a result, a "standard quotation" has yet to emerge. The basis for calculating costs differs from vendor to vendor.

Introduction to Discovery (1/7): Don't leave the selection of a discovery vendor to others! 　As mentioned in the previous section, each vendor has a different method of calculating whether to charge by the volume of data before or after decompression, and the best example of this is a case where the data volume expands several times after decompression, even though the unit price is low.

Do the legal staff on the side requesting discovery have such an understanding? It is dangerous to compare only the unit price if it simply says "hosting. Because this is an industry without clear guidelines, it is necessary to look at the total cost, not the unit price.

Recently, more and more companies are leaving the data they have collected hosted as it is, even after discovery has been completed. This is because the same data may be used in discovery for other litigation or investigations if the litigation or investigation develops into a cross-border matter. Therefore, companies that are likely to host the data on an ongoing basis may want to ask their vendors for hosting fees after discovery is completed.

(5) Project Management

In most international litigation discovery cases, it is advisable to deploy about two project managers to oversee the project. The project managers' roles include working with the company's legal staff to set search keywords, preparing batches of documents for review, and acting as a liaison between the company and the law firm to ensure smooth project progress.

In FRONTEO's experience, a project manager spends approximately 20 hours per month on a single project. However, please note that related travel and business expenses will be billed separately, and these costs will be incurred if the project is outsourced to an overseas vendor.

Example 2) Project Management Quotation

(6) Review Costs

It is no exaggeration to say that "Discovery Costs = Review Costs," as a large percentage of the costs will be review costs. As you can see from the estimates, most of the review costs are labor costs.

Therefore, an increasing number of companies are introducing automated computer analysis called "Predictive Coding" for reviews.

At FRONTEO, we have developed a predirective coding system that uses artificial intelligence to analyze Asian languages, and when we used it in an actual project, we were able to complete the review in one-third the time expected by other companies, compared to the review work performed by an American. In addition, since no labor costs were incurred, the cost was reduced to one-fifth of the original cost, which was highly evaluated by the client.

Whether or not to use predirective coding depends on the vendor's decision, and the accuracy varies. When reviewing quotes, it is strongly recommended that you check the Precision Rate and Recall Rate for Predictive Coding, in addition to whether or not Predictive Coding is used.

Incidentally, many discovery vendors are reluctant to provide these rates. Although they should have samples of data with a proven track record in discovery, it is a fact that in many cases, they do not provide the rates, using such evasions as "Japanese is difficult to use, so I can't provide the rates. This is because rates are directly related to review costs.

(7) Production

Production is the process of submitting evidence. Once the documents to be used as evidence have been selected through the review, the attorney representing the litigant decides which evidence files to submit to the opposing party. The vendor processes the data specified by the attorney into a file format (Tiff format) that is reliable as evidence, and if necessary, prepares an English translation of the file.

Depending on the case, roughly 3% of the files to be reviewed are eligible for production. If there are 90,000 files to review, 2,700 of them will be produced.

Previous：Introduction to Discovery (5/7): Discovery Vendor Technical Competence as Important as Discovery Know-How

Next Article: Introduction to Discovery (6/7): The Key to Controlling Costs is in the Estimate Check (Part 2)