An Electronic Customer Statement Delivery & Data Mining Solution Gives Merrill Lynch a Competitive Advantage in the Brokerage Industry

By Mason Grigsby, IMERGE Consulting


“Show me the statement” – with 1,200,000,000 Statements and Trade Confirmations electronically available in the newly installed RecordsNetSM system, Merrill Lynch clients can easily and conveniently obtain copies of their historical financial records. Designed to archive regulated information and allow Merrill Lynch business and support areas to quickly respond to customer requests for information and re-prints of documents – the RecordsNetSM system, with one year on-line today, will grow to a total exceeding 12,000,000,000 pages of computer data. This represents ten years of on-line electronic history. Easily the largest electronic repository in the Financial Services industry dedicated to providing access to statement and confirmation history.

Want to know the price of a stock that you purchased ten years ago? Given the ongoing accumulation of RecordsNetSM historical data Merrill Lynch will tell you – not only will they tell you, they will let you make your own inquiry over the Internet. Want a copy of your statement from 1 year ago, 2 years or 10 years ago – the RecordsNetSM system will be at your service. Merrill Lynch is in the process of building the history file, and when completed your Statements, Trade Confirmations, and year-end 1099 Interest and Dividend notices will be there – day or night 24 hours, 7 days a week.

Originally conceived in 1994, the Merrill Lynch Records Management Group, headed by Director Michael McCabe, initiated an investigation to ascertain the technology required to electronically store, index, access and re-print, exact replicas of private client, institutional, international and employee statements, confirmations and internal transaction reports. This global electronic document warehouse was envisioned as the worldwide corporate memory for all regulated customer and business data. The primary goal was to allow internal user groups to respond to customer requests for information or copies of documents quickly and easily. The development team also knew that an electronic data warehouse with a full ten-year history would create significant customer service differentiation for Merrill Lynch. What the team didn’t know was the perseverance that would be required to implement their vision.


The existing system of storing rapidly increasing volumes of statement and report documents on a combination of paper printouts and computer output microfiche was costly to produce and deliver to the appropriate business and support areas. The associated manual access and re-print process was also creating excessive cost and contributing to an increase in request fulfillment turnaround times. The long period of time required to obtain copies of statements, particularly to fulfill customer requests, was considered unacceptable.

The RecordsNetSM Enterprise Report Management system went live in April of 1998; however, there were significant technology challenges to overcome in the 4-year implementation process. The software to handle the indexing and retrieval of 100 million pages per month did not exist (Merrill knew that this would be one of the highest volume Enterprise Report Management systems in the world). The ability to keep that many documents on – line for ten years, while meeting the Securities and Exchange regulations that require storing statements and confirmations on non-erasable optical disk media, was also in question due to the sheer volume of documents. The problem was such that Merrill Lynch’s Records Management and Systems groups, which started with a list of twenty potential vendors, would be forced to look beyond the solutions then on the market and find an integrator that had the vision and creativity to “think out of the box.” The architecture and software would need to process, index, store and access a volume that would eventually grow to more than 12 billion pages. In addition, the system needed to insure individual Statement and Confirmation security – account, tax number and broker identification, that would allow both internal network and future Internet access without fear of compromise.

Merrill Lynch, in the course of business, maintains two IBM/MVS mainframe data centers, which produce all of the statement, confirmation and transaction report output. Since this data represents the primary customer information source, it was determined that the RecordsNetSM system would also need to be “mirrored” to insure that customer service would not be interrupted. The creation of duplicate index servers and optical media of the 100 million page per month output with the associated account level and tax identification index record was another major architecture issue that had to be resolved. These volumes mandated a recording (index extraction from the pages and update of the database) specification of a minimu of 1.0 million pages per hour to insure timely processing of month-end and year-end statements.

Both storing and viewing the Statements and Confirmations in their original IBM Advanced Function Presentation print format was considered highly important due to regulatory requirements. Of equal importance, however, was the desire to facilitate dialog with a customer and re-printing of the exact replica of the document. Given these requirements, the system architecture needed to be robust enough to index and store the original document format as well as provide concurrent access to the legal and compliance departments, the four customer service locations, and eventually more than 20,000 U.S. and international Financial Consultants.

The mandated Securities and Exchange storage requirements for statements and confirmation notices created the need for still another innovative software solution. Since the documents would be written on unalterable optical disks, they also needed to be stored on disks by document type. The indiscriminate writing of output to different non-erasable optical disks as it was produced by the MVS mainframes would complicate legal retention and compromise recordkeeping rules for broker- dealer documents. The software therefore needed to be capable of recognizing the different document types and only allow like documents with like retention periods to be written on any given optical disk.

The RecordsNetSM system, when fully implemented would represent a ten-year history of the corporate database. The development team therefore decided that it would be logical to enable users to search the document warehouse for specific information that would have high “business intelligence” value. The system specification therefore included high-performance software analysis capability – known as “Report Mining” that was capable of extracting specific field information from statements and trade confirmations, to create new “ad-hoc” information reports. The key element of this software was that the extraction process had to provide an easy to use ability to extract fields without programming – in effect a “point and click” process that any authorized employee could use. This again represented a requirement that would be a significant challenge due to the Merrill Lynch volumes.


Merrill selected a client/server architecture that would transfer the processing and retrieval of customer documents from the IBM/MVS mainframe and alleviate contention for processing. The new “mirrored” middleware application would be connected to the existing MVS systems using an ESCON channel to achieve the rates necessary to transfer the 100 million pages per month to the servers.

The “object oriented” index (the index is managed separately – creating pointers to the documents) is created and managed by an Oracle database that is highly segmented by the vendor to achieve the required processing speeds of more than one million indexed pages per hour. The significance of this task cannot be underestimated. There are few database architectures that have the ability to update index records at this sustained speed, considering that there can be no deletions to the file and update and retrieval rates to the document history cannot degrade from the required 1.0mm pages per hour as the file grows.

The RecordsNetSM system (figure 1), developed by Alysis Technologies of Emeryville, California, consists of two separate sites. Each site has five Sun Ultra II servers (each with 45 GB’s of storage) and one Sun E4000 server. Each Sun server has 250 gigabytes of magnetic storage that process and store index records. The design at each location utilizes multiple servers to simultaneously extract and update the index records: documents are recorded on 12″ (12.0 gigabytes per disk) optical disks manufactured by Phillips, LMS and maintained in a total of two (at each site) Cygnet, Inc. optical libraries, index records are then migrated and merged with a primary “retrieval” index server. The problem of storing that much data was greatly alleviated by the document compression that CommPres, Inc. developed – 5 to 1 for Statement compression and 20 to 1 for Transaction Report compression. The Merrill Lynch requirement for uninterrupted customer service meant that duplicate optical disks need to be created each night and sent to the alternate location for backup and “hot-site” availability. Index records are also written to optical disks during this process to insure that there is backup for the magnetic disk and to meet the regulatory recordkeeping requirements for the security and integrity of customer records.


Report Mining is an important element of RecordsNetSM since it allows access to “clean” statement and trade confirmation data for business intelligence. For example, when a client questions a specific account transaction, a Merrill analyst can automatically search and extract data from multiple statements and/or trade confirmations over a given time period and locate the transaction that is in question. The only other alternative to this information gathering process before RecordsNet was a manual “eyeball” of the document(s) in question. In addition, Merrill analysts will be able to prepare customized client reports by searching statements over time periods to analyze transaction trends and aggregate gains or losses.

The report mining software with its innovative decision analysis capability, allows Merrill analysts to design templates to extract specific information contained within a customer statement and/or trade confirmation. A query of the Statement or Confirmation history will enable Merrill Lynch to give clients replies to queries about stock purchases – the prices paid, the dates of trades, etc. for analysis.

Another example of Report Mining is illustrated by one of the brokerage industry regulatory filing requirements. An “Abandoned Property” report is mandated by each state, which requires

businesses to detail all of the accounts and assets that are unclaimed by clients. This manual process was extremely time-consuming. The “Report Mining” software with its’ innovative decision analysis capability, allows Merrill analysts to query fields on the statement. The data is automatically extracted from the statement and the assets in the account are totaled. This process, normally associated with a very sophisticated “Data Warehouse & Data Mining” system, is now easily performed without any programming – as a by-product of the existing Statement Warehouse.

John Ortiz, Vice President of the Operations Services Group for Records Management at Merrill Lynch states that, “as we continue to build the history file over the next ten years we will have the ability to quickly respond to client requests for copies of Statements and Confirmation Notices and provide unique data analysis capability.”

According to Lucia Palmeri, Vice President of the Technology Group, “the process of extracting fields for indexing and creating new document formats literally takes seconds”. Figure 2 illustrates the process of highlighting and extracting field data from the IBM Advanced Function Presentation (AFP) statement format.

Ortiz also stated that “the key to what will eventually prove to be a highly competitive tool is the ten-year permanent history. The document repository will not only give our clients access to their historical statements but allow them to use our Report Mining software to perform analysis of their own portfolios.”

The next process improvement on the horizon for the RecordsNetSM system includes the Internet as a future business portal. This phase will create worldwide access to reports, statements and confirmations. In addition, the year-end Tax Reporting Statement (1099 Interest and Dividend Notices) will be incorporated in the history file, thereby giving authorized users anywhere in the organization access to a complete “virtual” electronic client file folder.


If business and technology groups in any organization are to become strategic players, they must demonstrate that the Information Technology they are implementing has a very real impact in achieving the business goals of the corporation. Philip J. Villani, First Vice President, Operations, Services Group stated “RecordsNet is an example of Merrill Lynch’s commitment to investing in value-added information technologies. RecordsNet was developed by a dedicated team of business and systems professionals working in partnership with innovative software and hardware vendors.” The RecordsNetSM technology initiative has not only impacted front line customer service by improving the client-related business process, it did so with a dollar return that was well within the Merrill Lynch return on investment guidelines. Key areas of cost benefit included savings from eliminating microfiche production costs, reducing labor costs associated with Statement and Confirmation retrieval, and eliminating two in-house computer output microfiche processors. The “soft dollar” benefits of creating ad-hoc reports from existing reports and using Statements for business intelligence, while not a part of the return on investment analysis, were clearly key benefits in enhancing the ability of the organization to provide personalized customer service. As Michael McCabe stated, “Our primary goal was to insure that Merrill Lynch created a technologically advanced customer service system that would give us a competitive advantage. RecordsNetSM will become a global virtual electronic library that will provide access to all client financial information – anytime, anywhere in the world.”

The use of technology to serve more of the client base electronically, retain customer loyalty and attract new clients are integral components of the Merrill Lynch corporate strategy – the RecordsNetSM system certainly meets those criteria.

1999 Mason Grigsby, IMERGE Consulting, 2033 Leavenworth St.
San Francisco, CA 94133, Tel: 415-775-4282,
Email: info[at]alysis.com

toggle off
Gokreclame uitschakelen.