Market Overview_information Quality 2002

© 2002 Giga Information Group Copyright and Material Usage Guidelines

September 13, 2002

Market Overview 2002: Information Quality Lou Agosta Contributing Analysts: Keith Gile, Erin Kinikin, Philip Russom

Giga Position Giga last reported on the data quality market in August 2000. Since then, the total market has grown from $250 million to an estimated $580 million, including both software and services. The represents a compound annual growth rate (CAGR) of 66 percent, which is no small accomplishment in the past two years, albeit for a market of modest size. That growth consists of two parts — software and services — which have grown at rates of 40 percent and nearly 55 percent (respectively) during that period. Our prediction is that the data and information quality market will slow, with the respective growth rates reversed to 25 percent for software and 20 percent for services during the next year. The economy is too uncertain to predict beyond that at this time. At that rate, the market will hit $1 billion in 2005. The reason for this reversal of emphasis of software and services is a growing appreciation from IT managers that the information quality process requires software to automate, leverage productivity and manage what is otherwise a labor-intensive task. If data quality software vendors are able to help enterprises move beyond defect inspection to root cause analysis and intelligent information integration, then the huge potential of this market in improving the quality of enterprise information may indeed be realized much sooner than 2005.

Proof/Notes Information quality initiatives and the market for supporting technology are being driven by: 1.

Customer relationship management (CRM) implementations

2.

Public sector initiatives

3.

Mergers and acquisitions

4.

Data and information aggregation and validation

5.

Service bureaus/application service providers (ASPs)

6.

Data profiling

7.

Data warehousing and extraction, transformation and loading (ETL) operations

8.

Run-to-run balancing and control

9.

Profitable postal niches

10. Information quality features: reducing uncertainty via quality at no extra charge End-user enterprises will not be able to “plug into” data and information quality as easily as buying a software connector or adapter for an ETL tool or a CRM application. Therefore, enterprises should plan on completing design work to understand and reconcile the diverse schemas that represent customers, products and other key data dimensions in their firms. A design for a consistent and unified view of customers, products and related entities is on the critical path of successful intelligent information integration (see table at end of document).

Planning Assumption ♦ Market Overview 2002: Information Quality RPA-092002-00008 © 2002 Giga Information Group All rights reserved. Reproduction or redistribution in any form without the prior permission of Giga Information Group is expressly prohibited. This information is provided on an “as is” basis and without express or implied warranties. Although this information is believed to be accurate at the time of publication, Giga Information Group cannot and does not warrant the accuracy, completeness or suitability of this information or that the information is correct.

Market Overview 2002: Information Quality ♦ Lou Agosta

CRM implementations: Information quality issues have turned out to be the weak underbelly of CRM implementations. The need to address the data quality risk to CRM systems will continue to stimulate the acquisition of information quality software and solutions by end-user enterprises. Without information quality, the risk is that the client will implement CRM but miss the customer. For example, knowing the lifetime value of a customer still requires aggregating a lifetime of customer transactions across multiple systems and datasets, some of which are not available. Such intelligent information integration is something no mere user interface, no matter how elegant, can provide. Best-of-breed metadata administration, mechanisms to share designs as well as the data standardization tools from the data quality vendors, will be essential in helping technology catch up with the much hyped 360-degree view of the customer. Given the many valid and legitimate definitions of customer, this entity often ends up being a point on the horizon toward which service, delivery and marketing processes converge. There is no easy substitute for a piece of design work for data integration that optimizes transactional systems by closing the loop back from decisionsupport systems, preserving the business context of the organization in question. CRM has brought to the fore the task of identifying and deduplicating individual customers across multiple datasets, touchpoints and user interfaces. Much of the work in the data quality arena goes into accurately identifying customers as individuals, then building the business rules that warrant the outcome of a match or not. This requires data standardization, diagnosis of data defects, duplicate detection, error correction, resolution recommendations, approvals and notifications and ongoing monitoring. As Giga research has pointed out, enterprises should establish an information quality plan as part of any new CRM initiative and press CRM vendors to articulate a consistent, end-to-end strategy to identify and address customer data quality issues as an integral part of the customer process (see References). All of the best-of-breed data quality vendors in the table at the end of the document — Ascential, DataFlux, Firstlogic, Innovative Systems, Group 1 Software and Trillium Software — provide standardization, identification, matching and deduplication software that can be applied to a wide variety of data types, not just customer data. However, with the amount of activity surrounding CRM systems, vendors have been attracted to the market and positioned software solutions to address the real needs (see Planning Assumption, Market Overview 2002: Customer Relationship Management, Erin Kinikin) — it’s where the data is. For example, several vendors have adapters or connectors that specifically target Siebel: •=

Ascential INTEGRITY for Siebel eBusiness Applications

•=

Group 1 Data Quality Connector for Siebel

•=

Innovative i/Lytis for Siebel

•=

Firstlogic Siebel Connector

•=

Trillium Connector for Siebel (both versions 6 and 7)

Other data quality providers such as DataFlux and Sagent (GeoStan) do not yet have a separately predefined adapter for Siebel but support a software development kit (SDK) that can be used to jump-start the custom software development desk. Without exception, the challenge is to move beyond name and address standardization and attain intelligent information integration, something no single-system plug-in can provide. The installation of a new CRM system often acts as a lightning rod for the constellation of data quality issues as the system is loaded with data from other back-end systems in the enterprise. Often times, the piece of design work needed to identify the customer of records is the responsibility of the IT department at the enduser firm. Assuming that that department has been diligent, it is useful to have technology with which to implement the design rather than reinvent the wheel. This is where a prepackaged adapter or connector can provide a boost to developer productivity, provided that the inputs are understood and qualified. Public sector initiatives: Giga has recently heard from several information quality software vendors with

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 2 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

special capabilities in immigration and law enforcement. For example, in 1986 Australia was trying to determine how many illegal immigrants were in the country. Since Australia is an island continent, the government figured it would match the names of people who arrived for vacations, etc. with those who left, and the difference would be the number of illegal immigrants. However, out of a population of about 12 million, the number that was derived — 400,000 — was too high. Tourists with no knowledge of English or the alphabet — for example, tourists from Asia — did not necessarily know how to write their names and might transliterate their names differently after they had been in the country for several weeks or months. Thus, out of the crucible of an international melting pot, Search Software (renamed Search Software America) was created. The actual number of illegal immigrants was 100,000. Innovative Systems has a timely offering to the burgeoning security and law enforcement markets: i/Lytics Secure. This offering enables public and private organizations to compare client records against lists of criminal suspects and fugitives published by the US Treasury Department’s Office of Foreign Assets Control (OFAC compliance) and other leading US and international government agencies, such as the FBI’s most wanted fugitives list, the US Marshals’ most wanted fugitives, the US Drug Enforcement Administration’s fugitives list and Interpol’s wanted notices. Trillium Software reportedly offered to its government and financial customers a free service engagement to develop a process to identify and screen for OFAC compliance. Some took advantage of the offer. Others reportedly used the tunable matching technology already available within the Trillium Software System to identify and link suspect names and addresses. Developers and policy makers are advised to be skeptical about the appropriateness of high-risk CRM technology applications designed to manage consumer or business PC sales (see IdeaByte, CRM Takes on Terrorism, but Only Offers a Partial Solution, John Ragsdale). However, sophisticated matching technology, especially from the data quality vendors, properly tuned and backed up with the right databases and management policies, are likely to be an important part of the solution. The market for addressing information quality defects is being extended to improving aging database infrastructure and outmoded data management practices in the public sector. Thus, all types of databases in the public sector (for example, voter registration, driver’s licenses, hazardous waste haulers, pilot’s licenses, immigration, education and health records) will be enhanced and interconnected where appropriate policies and protections warrant it. From a public policy point of view, something such as national identity cards are far less significant than operational excellence and information quality work in the trenches, for instance, getting driver’s license and voter registration data stores accurately aligned and synchronized. Mergers and acquisitions: Mergers and acquisitions continue apace, and as soon as enterprises formalize a merger, the issue of compatibility between IT systems arises. In defining the new technology infrastructure, identifying, assessing and managing the overlap between customers, products and other essential data dimensions become a priority. The consolidation of product and customer dimensions enables cross-selling and upselling in CRM, presenting a single product catalog to partners and clients over the Web, as well as substituting information for inventory in demand planning and supply chain applications. A profile of the data and of the respective workloads surfaces a whole hierarchy of information quality issues. No reason exists why systems from completely different enterprises should be consistent or aligned or should satisfy a unified design. As a result of the merger, the two or more firms that are now, as a matter of definition, part of a single business enterprise risk an information quality meltdown unless the data is inventoried, evaluated and managed proactively as an enterprise asset. For example, client references from Firstlogic, Group 1, Trillium and Vality have reported the use of searching and matching technologies to produce customer-focused consolidated systems where previously heterogeneous, multi-divisional line of business or product-focused organizations had existed. Data and information aggregation and validation: The technical and business processes for data (and information) integration are so similar to those for data validation that it is hard to justify excluding such

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 3 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

matters from the information quality market. Credit bureaus and information aggregators have found that customer data integration and information quality is a growth industry. After having built up databases of individuals and households in the 100 million record range as a result of consumer credit and credit reporting, firms such as Equifax, Experian and TransUnion now find they are in the business of verifying, identifying and integrating customer data. Database marketing and demographic firms such as Acxiom, CACI, Claritas, Harte-Hanks and Polk see similar opportunities. For example, Acxiom reportedly has more information on Americans than the Internal Revenue Service (IRS) does. Acxiom acquired DataQuick, one of the largest compilers and resellers of real estate information in the country, and it can identify home ownership. By using Acxiom’s AbiliTec customer data integration technology, the end users of customer loyalty (and related) applications can link contact information captured in real time across multiple customer interfaces with legacy customer data stores. Experian claims that its Truevue CDI solution adds a significant lift to the matching and deduplication results by checking a reference database with some 205 million entries for US residents. This market segment is highly competitive and alternative vendors are discussed in previous research (see IdeaByte, Information Infomediaries for Demographics, Lou Agosta). In most instances, enterprises must use data quality functionality narrowly defined to profile, standardize and link their customer data before matching to any of the above cited external databases. Service bureaus/ASPs: The ASP market has not fared well since August 2000 when Giga reported that several data quality vendors were extending their solutions in the direction of the Web, with an emphasis on eCRM ASPs. At the time, Firstlogic, Group 1, Innovative Systems, Trillium and Vality (whose INTEGRITY is now part of Ascential’s enterprise data integration solution as well as a stand-alone) were all working on separate versions of an ASP offering to allow clients to process individual records on a transaction-bytransaction basis (which yields real-time data quality). The revenue model is a vendor’s dream and can be cost-effective for clients with high-value, low-volume transactions. The problem has been forecasting revenue using a “by the transaction approach.” The service bureau has a similar tradition, without the emphasis on specific application functionality, and that is the direction this submarket has taken. For example, Firstlogic has exposed its data assessment functions via a service offering with reporting over the Web via IQ Insight; in addition, Group 1 has Hot Data, Sagent has Centrus Record Processing and Acxiom has Acxiom Data Network (see IdeaByte, Uses and Limitations of Real-Time Data Quality Functions, Philip Russom). Data profiling: Data profiling products and services include Ascential MetaRecon (available either as a stand-alone or as part of a data integration suite), Avellino Discovery (also available from Innovative Systems), Evoke Axio, Firstlogic IQ Insight, and Trillium Software’s Data Quality Analytics. Data quality should be checked every time the data is touched, at the front end, middle and back end of the system. This applies first to the operational transactional systems. Data entry screens, user interfaces or points of data capture should include validation and editing logic. Of course, this implies the rules of validation and the relations between data elements are known and can be coded. As indicated, this guideline extends to any and all systems, operational or decision support (data warehouse). The days are gone when Evoke was the only technology in the data profiling market and its prices reflected it. The competition between Ascential MetaRecon, Avellino Discovery (also available from Innovative Systems) and Evoke Axio will cause prices to fall further, which is good news for buyers. Data warehousing and ETL operations: The architectural point at which upstream transactional systems are mapped to the data warehouse is a natural, architectural choke point through which all data must flow in the process of extraction and transformation on its way from the operational to the decision-support system. It is a natural and convenient place to perform a number of essential functions, such as checking and correcting information quality and capturing metadata statistics about information quality. The process is often automated by means of an ETL tool, so it makes sense to look for connectors from the ETL to the data quality technologies. The available options are: •=

Firstlogic Information Quality Suite can be invoked by ETL technology from Ascential (DataStage) and Informatica (PowerCenter).

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 4 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

•=

Ascential INTEGRITY can be invoked by Ascential (DataStage) and, at least prior to its acquisition by Ascential, by Informatica (PowerCenter). Of course, Ascential is promoting the advantages of getting both data quality and ETL technology from the same provider — in this case, Ascential.

•=

Trillium Software System can be invoked by Ascential (DataStage), Informatica (PowerCenter), Microsoft DTS, Ab-Initio and, at least prior to its acquisition by Business Objects, by Acta.

•=

DataFlux Blue Fusion is integrated with SAS Warehouse Administrator.

•=

In November 1999, Oracle purchased Carleton and its Pure Integrate contains data standardization technology. However, according to Oracle, Pure Integrate is being desupported as a stand-alone tool. Oracle Warehouse Builder (OWB) 9i includes the Name/Address operator of Pure Integrate and full support of the product functionality should be available by 2003. Meanwhile, Oracle reports that technology from the Trillium Software System is being integrated into OWB.

•=

Group 1 DataSight is able to invoke Informatica and iWay, thus turning the tables on the ETL tool and subordinating the ETL process to the data quality one.

Note that, for once, the promise of a “seamless interface” is not mere marketing hype. In demonstrations provided to Giga, the integration between Ascential and Informatica and the respective data quality technologies were such that the data quality features appear as subfunctions of the ETL process and not as part of a separate window or tool. According to DataFlux, a similar consideration applies to it and SAS Warehouse Administrator. Run-to-run balancing and control: Data center operations are highly complex in diverse ways. The output of one process is the input to another. Many-to-many relationships between jobs and processes occur in intricate patterns. Mounting the wrong input or mounting a duplicate input (whether tape or other media) can wreck havoc with downstream jobs. Even if the proper input is used, the record counts can be inaccurate. Software to perform run-to-run balancing and control based on the particulars of the application has been available on the mainframe for many years and has now migrated to alternative platforms, including Unix in its many forms and NT/W2K. Vendors such as Unitech Systems have been doing business since the mid-1980s and have kept pace with the technology innovations in platforms and processes. This is definitely a problem space of the utmost importance; it is deeply embedded in the operational matrix and addresses risks to data quality from the perspective of operations. Vendors that compete with Unitech and offer competing run-to-run balancing solutions include Accurate Software, Smart Stream and Checkfree. Profitable postal niches: For firms requiring direct mail, the deliverable is a standard name and address that can be printed on an envelope. Firstlogic offers PostalSoft, Group 1 offers a name and address information system (CODE-1 Plus), Ascential (Vality) offers INTEGRITY CASS Appended Information Modules (AIM). CASS is the coding accuracy support system, certified by the US Postal Service to ensure accuracy, completeness, consistency and validity of US address information in the corporate database. According to DataFlux, with the dfPower Address Verification module, addresses can be verified directly within the database. This is a competitive market with significant barriers to entry due to the complex rules for postal name and address formatting and the diversity of foreign and local names and noise words, such as “Dr.,” “For Benefit of” or “In Trust for.” Companies such as Quick Address (QAS.com), AND Group and Human Inference offer solutions for international name and addresses in the European Union (EU). Group 1 claims it understands the postal formats of some 220 countries, works with the International Postal Union (IPU) to keep the address lists current and lines up with catalog merchants in using address-based identifiers data as the way to isolate households. Group 1 is able to append demographic data in any legally permitted detail and offer a service bureau for address checking on a transaction basis. This eliminates the need for end-user firms to update the 220 databases of country addresses. Name and address preparation for the US Postal Service (or any national postal system) is a dull application that costs a small fortune if a firm makes an error. Thus, a specialized application is important here. Information quality future — reducing uncertainty via quality at no extra charge: Information defects

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 5 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

generate uncertainty in simple ways, such as when the address is wrong and the client’s package is lost. They also cause uncertainty in complex ways, such as when what counts as a “sale” is ambiguous between receiving a customer order vs. receiving a customer payment. Thus, the need for business enterprises to manage and reduce uncertainty drives forward the information quality imperative. This imperative addresses and, if successful, reduces the risk of surprises such as schedule slips, disappointed customers, solving the wrong problem or budget overruns due to information quality defects. This, in turn, enables implementation of state-of-the-art data warehouses that support high-impact business applications in CRM, supply chain management and business intelligence. Due to a wide diversity of heterogeneous data sources in the enterprise, both design and implementation skills will be needed to transform dumb data into useful information. When the costs of defective data are added up — operational costs such as job failures, rework and duplicate data, as well as missed business opportunities due to uncertainty — information quality is available at no extra charge. Giga predicts that technology for deduplication, customer identification and entry-level data standardization functions will ship “at no extra charge” with the underlying relational database by the end of 2003 [.7p]. This is analogous to the trend whereby ETL technology, data mining and online analytical processing (OLAP) have been driven into the relational database.

Alternative View Two factors will limit the growth of the data quality market to single digits: The first relates to vendors, the second to end-user firms (customers). First, a review of the software features and functions indicates many data quality software providers are technology driven, not market oriented. The business application is, by definition, where the business value shows up. But most of the data quality firms seem constitutionally lacking in the ability or experience to market data quality as an enabler at the application level. Another constraint that will significantly limit the growth of the data quality market is the similarity between the corporate reaction to data defects and the discovery that a firm has been the victim of fraud. Firms do not want the events discussed publicly. It seems to imply lack of diligence, skill or intelligence on the part of the victim. Whether data quality is the weak underbelly of CRM (and diverse other applications) is a fact that will remain a well-kept secret, because lack of data quality opens up potential liabilities that firms have an incentive to overlook and not mention. Absent an information quality safe harbor policy to drive out the fear of reporting data quality issues, such issues will continue to be swept under the rug. Thus, the prospective end users of data quality products will continue to be addicted to data and drowning in it, all the while living in denial of data quality issues.

Findings Since August 2000, the total market for data and information quality has grown from $250 million to an estimated $580 million, including both software and services. The represents a CAGR of 66 percent, which is no small accomplishment during the past two years, albeit for a market of modest size. That growth consists of two parts — software and services — that have grown at rates of 40 percent and nearly 55 percent (respectively) over that period (see table below). All of the best-of-breed data quality vendors on the table of market shares — Ascential, DataFlux, Firstlogic, Innovative Systems, Group 1 and Trillium — provide standardization, identification, matching and deduplication software that can be applied to a wide variety of data types, not just customer data. However, with the amount of activity surrounding CRM, these software providers have been attracted to the market and have positioned products and solutions to address the real needs — it’s where the data is. The market for addressing information quality defects is being extended to improving aging database infrastructure and outmoded data management practices in the public sector. Thus, all sorts of databases in the public sector  voter registration, driver’s licenses, hazardous waste haulers, pilot’s licenses, immigration, education and health records  will be enhanced and interconnected where appropriate policies and protections warrant it. From a public policy point of view, something such as national identity cards are far less significant than operational excellence and information quality work in the trenches, for instance, getting driver’s license and voter registration data stores accurately aligned and synchronized.

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 6 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

Stand-alone data aggregators have found that customer data integration and information quality is a growth industry. After having built up databases of individuals and households in the 100 million record range as a result of direct marketing and consumer credit and credit reporting, firms such as Acxiom, CACI, Claritas, Equifax, Experian, Harte-Hanks, Polk and TransUnion now find themselves in the business of verifying, matching and integrating customer data. The point at which the ETL tool is invoked is a good time to check data and information quality. Data quality vendors that can be invoked from ETL tools are cited under Recommendations. When the costs of defective data are added up — operational costs such as job failures, rework and duplicate data, as well as missed business opportunities due to uncertainty — information quality is available at no extra charge. The prediction is that technology for deduplication, customer identification and entry-level data standardization functions will ship “at no extra charge” with the underlying relational database by the end of 2003. This is analogous to the trend whereby ETL technology, data mining and OLAP capabilities have been driven into the relational database.

Recommendations End-user enterprises will not be able to “plug into” data and information quality as easily as buying a software connector or adapter for an ETL tool or CRM application. Therefore, enterprises should plan on completing design work to understand and reconcile the diverse schemas that represent customers, products and other key data dimensions in their firms. A design for a consistent and unified view of customers, products and related entities is critical path of intelligent information integration. Certified data quality interfaces that specifically target Siebel are available — Ascential INTEGRITY for Siebel eBusiness Applications, Group 1 Data Quality Connector for Siebel, Innovative i/Lytics for Siebel, Firstlogic Siebel Connector and Trillium Connector for Siebel. Be prepared to perform additional design work, since no single-system plug-in can provide integration for multiple upstream data sources. End-user clients in the public sector will find that the data and information quality vendors are increasingly responsive to their special needs and willing to go the extra mile to accommodate procurement and bidding requirements. Innovative Systems, Search Software America (SSA) and Trillium have solutions targeting law enforcement and immigration applications. In particular, SSA has a long history of being able to tune its technology for “high-risk” applications where it is unacceptable to get a false negative (e.g., a criminal is allowed entry into the country). Group 1, SAS, Trillium and Ascential also have separate groups to handle government contracts. The competition between Ascential MetaRecon, Avellino Discovery and Evoke Axio will cause prices to fall further, which is good news for buyers. Organizations in need of data profiling and analysis tools should plan on exploiting these competitive market dynamics. Firstlogic IQ Insight exposes data assessment functions via a service bureau offering with reporting over the Web — Group 1 uses Hot Data, Sagent uses Centrus Record Processing and Acxiom uses Acxiom Data Network. Trillium Software System, INTEGRITY and Firstlogic can be invoked by ETL tools from Ascential and Informatica in a way that makes them look and behave like subfunctions of the ETL tool. So “seamless” is not hype — the integration is truly seamless. Likewise, SAS Warehouse Administrator is fully integrated with DataFlux in a similar way through the BlueFusion Client-Server implementation. Group 1 is able to invoke Informatica and iWay ETL technologies. For organizations requiring data quality processing using the mainframe, Trillium, Innovative Systems,

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 7 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

Group 1 and DataFlux (SAS) and Ascential (INTEGRITY) are first choices. SDKs that enable mainframe processing are also available from SSA and Innovative Systems. End-user clients with data quality issues that cannot be adequately addressed by standardization and parsing, high-performance requirements where standardization presents unacceptable overhead, high-risk applications (e.g., law enforcement or immigration), or multi-language or global naming requirements (including doublebyte character sets) will want to consider a combination of stabilization algorithms (e.g., Soundex, NYIIS) and context-based rules as exemplified in SSA’s and DataFlux’s approaches. Trillium Software is UNICODE enabled and has references that are cleansing double-byte character sets. It reportedly has more than 40 customers using the product on Japanese characters sets in Kana, Kanji, KataKana and Romanji. Data quality should be checked every time the data is touched, at the front end, middle and back end of the system. It is not a one-time event but a continuous process. This applies first to the operational transactional systems. Data entry screens, user interfaces or points of data capture should include validation and editing logic. Of course, this implies the rules of validation and the relations between data elements are known and can be coded. If not, an additional data engineering and profiling task is implied. As indicated, this guideline extends to any and all systems, operational or decision support (data warehouse).

References Related Giga Research Planning Assumptions Data Quality Methodologies and Technologies, Lou Agosta Developing an Integrated Customer Information System — Approaches and Trade-Offs, Richard Peynot and Henry Peyret IdeaBytes Customer Data Quality: The Newest CRM Application, Erin Kinikin CRM Is Not a Substitute for Good Customer Data Quality Processes, Erin Kinikin Information Quality Market Drivers, Lou Agosta The Public Sector Gets Serious About Data Infrastructure Improvement, Lou Agosta Data Quality Breakdown in Election Spells Opportunity for Technology Upgrades, Lou Agosta IT Trends 2002: Data Warehousing, Lou Agosta Search Software America Has Novel Approach to Overcoming Poor Data Quality, Lou Agosta Data Quality — It’s Not Just For Data Warehouses, Lou Agosta Look at Evoke’s Heritage to Grasp Its Special Capabilities, Lou Agosta Information Infomediaries for Demographics, Lou Agosta IT Projects Accessing Legacy Databases Benefit From Data Profiling During Planning, Philip Russom Uses and Limitations of Real-Time Data Quality Functions, Philip Russom

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 8 of 9

Market Overview 2002: Information Quality ♦ Lou Agosta

The Data Quality Market 2001 (in millions of dollars) Software Revenue

Service Revenue

Total SW & Service Revenue

Share of Total SW Revenue

Share of Total Service Revenue

Share of Total Revenue (SW & Service)

Group 1 (DataSight)

$36

$42

$78

14%

13%

13%

Firstlogic (Information Quality Suite)

$35

$22

$57

14%

7%

10%

Ascential (INTEGRITY, Metarecon, and DataCleanse DS XE)

$33

$15

$48

13%

5%

8%

Trillium Software System

$31

$4

$35

12%

1%

6%

QAS

$30

$3

$33

12%

1%

6%

Innovative Systems (i/Lytics)

$20

$5

$25

8%

2%

4%

Evoke

$8

$2

$10

3%

1%

2%

DataFlux (BlueFusion) $6

$1

$7

2%

0.3%

1%

Avellino (Discovery)

$4

$3

$7

2%

1%

1%

Data aggregators/ integrators

$0

$210

$210

0%

63%

36%

Other

$46

$24

$70

18%

7%

12%

Total

$249

$331

$580

100%

100%

100%

Source: Giga Information Group

Planning Assumption ♦ RPA-092002-00008 ♦ www.gigaWeb.com © 2002 Giga Information Group Page 9 of 9