Market Insights' founders thought they could implement a sophisticated technology project--a data warehouse--themselves. Their optimism almost cost them the company

One spring day in 1997, Rick Louie got a phone call that would change not only his future but the entire direction of Market Insights, the health-care data-analysis company that he and his partner, Anthony Milano, had founded. On the other end of the line was Marc Bandt, who had once worked for a former client of Market Insights. Bandt presented an intriguing idea. He had recently helped a large hospital in Houston set up a data warehouse--a highly sophisticated computer database for storing and analyzing billions of bits of information. Bandt wondered whether he could team up with Louie and Milano to build a data warehouse for their company.

Louie couldn't believe his ears. Just one year earlier, he and Milano had been debating that very idea. For the previous four years Market Insights had been compiling and analyzing vast quantities of health-care data and selling those analyses, quite successfully, to its clients--hospitals and health-care consultants. But the founders knew that their business was on the cusp of real change. Paper was on the way out, digital files were in, and the Internet loomed. To take on more clients and service them more efficiently, Market Insights needed a faster path to its data.

"I knew we needed to move to something more sophisticated, some kind of technological change that could better position us to attack the marketplace," Milano recalls. "But I wasn't confident about which one was right yet." Still, he was convinced of one thing: what Market Insights needed was a supercharged, highly specialized--and highly costly--data warehouse to handle the massive amounts of health-care-industry data that it routinely processed.

His partner was less confident. While he valued Milano's opinion, Louie was uncomfortable spending hundreds of thousands of dollars on technology. Market Insights was a self-funded company that had turned a profit every year--and Louie planned to keep it that way. "We had both come to the decision that to grow, we had to become a technology company with an Internet presence," he says. "We were both in agreement for the vision. But some of the expenses, I wasn't excited about." And then came Marc Bandt's phone call.

The idea was sound enough: using a data warehouse, the company's analysts could more quickly and efficiently churn out the complex, custom benchmark reports that were Market Insights' specialty. But more important, the company's clients could be connected to the warehouse through the Internet and peer for themselves into the massive reserves of public data that Market Insights had stored there. The result? A do-it-yourself information service. In exchange for an annual subscription fee, clients could rustle up quick answers to questions about, say, the costs of hip-replacement surgery or emergency-room protocols.

Of course, clients would still need Market Insights to produce in-depth, full-blown reports. But when a simple analysis or a limited amount of information would do, buyers could get it cheaply and quickly through the new service. With the boom in health care, there was no telling who among insurers, researchers, students, journalists--you name it--might want to dive into Market Insights' pool of data if only they had easy access. For its onetime investment to build the system, Market Insights would reap recurring revenues--the Holy Grail of small business.

For his part, Milano felt he had found a kindred data-warehouse spirit in Bandt. Not only did Bandt understand data warehouses, he had built one as well. "Marc spoke in a polished language from working with others in the field, whereas I was the only one here who really knew the kind of database I envisioned," Milano recalls. "Sometimes when people tell you what you already know in a polished manner, it's very powerful."

Bandt's suggestion for a data-warehouse investment resonated with Louie, too, as it added coherence to the ideas that Market Insights had been batting around. Louie respected Bandt's industry experience, and he liked the fact that someone who had known Market Insights from the client side now wanted to join its ranks. What's more, after conferring with Bandt, Louie worried that if the company didn't move right away, it would lose its window of opportunity to be an industry leader in this area. "We saw the industry was moving to the Internet, and we didn't want to be in the position two years from now of saying, 'We need to be there," he says.

The three agreed that command of cutting-edge technology would boost Market Insights' corporate value if the founders were ever to sell the company or go public. "He [Bandt] said, 'Let's grow Market Insights and take it to the next level; to do it, you need to buy the technology and invest in the future," Louie recalls. "We thought it would be tight, but we could buy the equipment, hire Marc, and go for it." Besides, 1997 was turning out to be a good year for Market Insights. "When we brought on Marc, who said this was a worthwhile investment, I was swayed."

Today, just about two years later, Louie and Milano have poured nearly $1 million--not to mention buckets of sweat--into hardware, software, and resources. The project has swallowed up the business's cash and employees. And it has yet to return any real money.

How could this have happened to a company like Market Insights, a business with revenues projected at up to $2.4 million this year, one that's run by two smart guys with plenty of business and technology savvy? It's a cautionary tale, one to be heeded by any small company that is even considering bringing a substantial technology project in-house. A canyon-like chasm lies between off-the-shelf technology and a major custom implementation: experience with small, homegrown systems is no guarantee of success on a larger scale. Even Market Insights, despite its sophistication, fell victim to poor planning, bad advice, finicky technology, and unready clients.

Though he and Louie are resolutely moving Market Insights forward, the whole episode still embarrasses Milano. "I feel crazy telling you this," he says. "I feel like an idiot. I have to chalk it up to inexperience. Otherwise I'd have to chalk it up to stupidity."

Market Insights started out with a bang. In 1993, Louie and Milano--then ages 25 and 24, respectively--ditched lucrative consulting jobs for their own business and 300 square feet of cramped office space at the San Francisco Bay Area World Trade Center. Louie, a Harvard grad, had been working as a health-care consultant for American Practice Management (now part of CSC Consulting) in San Francisco. More than 2,000 miles away, in Cleveland, Milano was a quant jock for McKinsey & Co. They were introduced by Jung Shin, a mutual acquaintance, with whom Louie hatched the idea of producing analyses--in any industry--for time-strapped consulting firms. Shin lured Milano to San Francisco, where, backed by $15,000 in combined personal savings and armed with their 386 personal computers, the three opened Market Insights (now officially named Information Insights Inc.).

By the end of their first month, they had landed a $25,000 contract with a Big Six accounting firm to analyze local hospitals. A short time later the company began to sharpen its business focus and pursue clients in the data-starved health-care industry. (Shin left at the end of 1994 to go to work for a competitor.)

In an industry with scalpel-thin margins, hospitals and providers of managed health care are hungry for details that will boost their bottom lines. Not only do they crave a diagnosis of their own operational health, but they also want the lowdown on competitors.

Armed with the latest number-crunching capabilities, a generation of data providers has sprung up to help them. Some resell the details they glean from clipboard surveys of hospitals and other caregivers. Others, like Market Insights, buy up the troves of public databases that are for sale and submit them to rigorous analyses. Take the Department of Health and Human Services' (HHS) collection of data from Medicare providers. HHS requires any provider requesting reimbursement for even one Medicare patient to submit an exhaustive cost report. Pay HHS something like $75,000, and the collection of reports can be yours. State governments are another bounty. Many mandate that health-care providers pony up inpatient and outpatient data as well as everything from the number of hospital beds to food costs. Anyone with a checkbook can buy the data from the government agency that collects it. But it's a seller's market: the state of New York charges a bargain-basement $600 for its data, Pennsylvania a cool $30,000.

From the start, Market Insights put a consulting spin on its information. Most of its competitors crank out standard reports several hundred pages in length, delivered in three-inch-thick three-ring binders. But Market Insights' forte is to churn the numbers into custom analyses that are short, sweet, and powerful--35 pages of graph-supplemented text. "We're the Cliffs Notes of our industry," says Louie. "You can buy War and Peace, or you can buy the Cliffs Notes."

For example, a hospital eyeing new federal guidelines for outpatient-procedure reimbursements might want to see how the changes will affect its bottom line. A Market Insights analysis can not only project the possible bad news awaiting the hospital--that its reimbursements might be lowered, say--but also propose a way to remedy the problem by identifying the specific medical procedures on which it will lose money.

Louie and Milano have always run Market Insights with the meticulousness of two numbers guys. "We're driven by making money every month, building real product and making profits," says Milano. Their approach was successful. Since its founding Market Insights has turned a profit every year and grown 25% to 40% annually. And in June 1998 Health Forum Inc., a subsidiary of the $100-million American Hospital Association, agreed to comarket its products. "We liked their culture and their expertise," says Patty Riskind, Health Forum's executive director of business development. "They're R&D oriented and extremely customer focused. Very entrepreneurial. They're smart guys, and they like to have fun."

Fun isn't a word you often hear associated with health-care benchmarks. However, technoliterate is. All along, Market Insights has relied on its founders' technical muscle to manage the vast quantities of health-care data that are its lifeblood. The company's first big-ticket technology purchase was a $2,000 digital-audiotape (DAT) drive. As boxes of data arrived on tapes, the information was dumped into the drive. From there it would move through a cable to the hard drives on the company's PCs where it was analyzed, and then the raw data were sent back to the tape drive for storage. Later, for even greater capacity, the company compressed the data it bought on tape and stored the information on CD-ROMs in a jukebox that held 250 disks. Via the local area network, Market Insights' four full-time analysts pulled up data as they needed it onto their hard drives.

The system worked, but it wasn't enough. By 1996 the new offices to which Market Insights had moved just the year before (on Brannan Street, in the heart of San Francisco's Multimedia Gulch) were crammed with cardboard boxes full of tapes. Milano believed the answer to their data-management problems lay in technology--specifically, a massive and sophisticated relational database. Simple relational databases, like the ones used in billing and accounting systems, pull up one record at a time and perform some operation on it. That style of handling data is perfect for recording a customer payment, for example, or processing an employee paycheck. But since Market Insights uses its data in a much more complex fashion, making numerous comparisons and projections based on all sorts of criteria, it needed something more advanced. It needed to be able to pull up many records simultaneously and to manipulate the information in a variety of ways.

Enter the data warehouse. As the name implies, it's a big box filled with digital information. But it's a smart box, optimized to concurrently store and analyze billions of bits of information.

For Market Insights the leap from CD-ROM jukebox to data warehouse was the equivalent of converting their data housing from a pup tent into the Taj Mahal. At the most rudimentary level, the elements of a data warehouse--roomy information storage combined with some analytical capabilities--can be assembled using standard relational-database products. A beginner's data warehouse can be had for less than $100,000, built from Microsoft products like an Access database, an SQL Server, and a Windows NT operating system. Spend two and a half times that and you can get more storage and greater analytical power: an Oracle database and a data-access tool from a well-known maker like Business Objects or Cognos.

To Milano, who handles operations and technology for Market Insights, the rudimentary setups were no match for the vast reservoirs of data that his company processed. He was certain that only a highly specialized data warehouse could fill its needs. The decision, though, wasn't his alone to make. Louie and Milano have always worked as a team. "I've still got to convince my better half of the promise of something like this," he recalls thinking. When the more cautious Louie balked at such a large capital expense for technology, the idea was shelved--until a year later, when Marc Bandt called with his proposal.

With a handshake, Louie, Milano, and Bandt entered into a deal. "I was looking for a company that had a large client base to sell into," says Bandt. "I thought there was a tremendous opportunity to leverage the core analytical set." Louie and Milano, too, saw the data warehouse acting as a springboard for new products and services. They had no market research in hand, just hunches on how information use would blossom.

For starters, they planned to load much of the health-care information they had collected into the data warehouse's roomy quarters and then sell Web-based access to it. (They eventually dubbed this service Agent 24-7.) It seemed a natural way to offer more to current clients as well as to attract new ones. Even today, good old-fashioned paper reports generate 60% of the company's revenues. Fifteen percent of that comes from spiffing up and reselling raw data--that is, reformatting and presenting it in a ready-to-crunch manner--from some of the 50 public databases Market Insights purchases every year. (A typical buyer might be a medical-device manufacturer that wants to sift through the data to prospect for new customers.) Desktop software applications, the company's other product line, were launched about two years ago. Financial Compass, to name one example, lets a hospital's chief financial officer call up current cost reports from 6,000 U.S. hospitals in a spreadsheet and calculate performance benchmarks for his or her own facility based on expenses, debt-equity ratios, liquidity, and so on. With the warehouse, Market Insights could offer clients something different: a kind of drive-through data service. Instead of paying for an in-depth report or a massive database, customers could grab some quick information and limited analysis.

The trio also came up with plenty of other ideas for putting the data warehouse to use. For a subscriber fee, hospitals and caregivers might stash their own facilities' data in Market Insights' warehouse. Then they could compare their own stats with other providers'. Maybe they would pick through the data to understand which pneumonia treatments had the best and most cost-efficient outcomes. Even the analysts at Market Insights would benefit, by hooking in and generating information more quickly.

As for how the data warehouse would get built, Market Insights had always been a technology self-starter. The company employed just one full-time software developer, but Louie and Milano say Bandt advised them that Agent 24-7 could be built with Market Insights' existing staff. (Bandt himself recalls telling them that it could be built with existing staff or they could easily hire someone with the technical capabilities, since they were in San Francisco.) "We had assurances from Marc that it could be done based on what he would contribute and considering myself and Nino [Milano] as resources as well," Louie recalls. "That's been the critical success factor of Market Insights in the past: where it takes most companies x number of employees to get the job done, we prided ourselves on doing it with many fewer, because we were smarter."

They agreed that Market Insights would pay Bandt a modest salary and foot the bill for the equipment. In exchange, once the system was up and running and had reached several milestones, Bandt would take an equity position in the company. Bandt says he was clear from the start about his reasons for working at a drastically reduced salary. "I was willing to work for them to prove to them that I was an equal in their circle," he says. Then, after acquiring his equity stake, he was counting on its becoming more valuable, either through the sale of the company, an initial public offering, or even simple growth. That would produce a financial windfall for him as part owner. "I was looking for personal wealth," he says bluntly.

However, neither the details of the project nor Bandt's arrangement were written down. There was no schedule for deliverables or project progress, nor any outline of responsibilities. And little was budgeted for consultants' fees, which in large implementations like this one typically pile up to four times the cost of the software.

"That was a mistake," Louie says today. "Our commitment was the money, which was all of our profit for the year. His [Bandt's] was to come here from Texas, take a $40,000 salary, and drive the project. That was enough on both sides." As for a contract, says Louie, "we figured we'd eventually get around to it. We were all good people, and we had worked together before. He had been our client, and there was a level of trust. Something that seems so obvious and important became a secondary priority."

In July 1997, Bandt joined Market Insights as chief technology officer. Over the next few months the three began purchasing the data warehouse's components--all top-shelf. They chose Informix Corp.'s Red Brick Warehouse database for its ability to juggle the voluminous amounts of data Market Insights used. Bandt advised using a sophisticated software called an on-line analytical processing tool, or OLAP. This added a new dimension to Agent 24-7: unlike a cheaper processing tool, such as the Java programming language, OLAP permitted users not only to pluck static data from the database but also to analyze it, factoring in variables and producing a tailored set of information. Agent 24-7's number-crunching ability would set it apart in the marketplace. Of course, the sophistication came at a price. The Information Advantage Inc. DecisionSuite that Bandt recommended cost at least 10 times as much as a Java application. But given the advantages it promised, even the usually thrifty Louie agreed it was worth it.

The upscale theme continued with the operating system and the hardware. The data warehouse's horsepower demanded the robust and complex Unix operating system, not to mention reliable hardware to run it on. In December, Louie and Milano authorized the $350,000 purchase of the dual-processor Digital AlphaServer 4100 running Unix. They financed much of the cost in a lease-to-own deal with Digital that they secured personally. In addition, they shelled out $140,000 for licenses for the DecisionSuite software and another $90,000 for the Red Brick data warehouse. They paid for the software with a combination of cash and bank loans. Total damage: $580,000. And that was just the beginning of the project's costs.

The arrival of the new equipment at Market Insights' Brannan Street office in January 1998 was cause for celebration. But the party was quickly over. By February, it was clear that the project's manpower requirements had been drastically underestimated, but because of the warehouse's high costs, there was no cash available with which to remedy the problem.

A much deeper wrinkle developed in March, when key software supplier Information Advantage approached Louie and Milano: in exchange for helping Information Advantage create new software applications, the vendor would provide consulting services to Market Insights so that it could complete the data-warehouse project. Then, at the end of three years, Information Advantage would acquire Market Insights.

It wasn't that Louie and Milano hadn't thought about selling their company before. They had--lots of times. "Rick and I felt the money was out there for the taking," says Milano. "The question was, what do you have to give up to get it, and what do you get besides the money?" The answers had always been "too much" and "not enough."

The offer was tantalizing but for all the wrong reasons: if they were going to sell Market Insights, they wanted it to be a strategic move to grow the business that they had nurtured for so long. They didn't want it to be because they needed cash. To make matters worse, Bandt was in favor of the deal, and that bothered Louie and Milano. "It was a strange dynamic," Louie says. "From Marc's point of view, it was a no-brainer. But we felt misled. We brought him in with all this money and the understanding that the sweat equity could make it a success. Now he was telling us that we had to sell off part of our company to make it [happen]." He and Milano were willing to consider Information Advantage's offer, Louie recalls, "but it was hard to stomach."

Bandt says the Information Advantage buyout "seemed like it might be an OK idea." Louie and Milano were open to all sorts of funding options, from buyouts to venture-capital investors, he notes. "I wasn't dragging them kicking and screaming."

Louie and Milano decided against the offer, resolving to complete the data-warehouse project on their own and get the company back on its feet. In April they allocated their only staff developer to the project full-time. Louie and Milano, too, began devoting additional hours to the data warehouse and to the key task of designing how the data was to be loaded into it.

But by the fall Market Insights was falling further behind. Milano widens his eyes when he recalls the gulf between expectations and reality. "He [Bandt] said we'd have something up in a month, and here it is like nine months down the road and we don't have anything. You know what I mean? It was amazing."

Bandt, who says that six, not nine, months had passed, defends the project's progress in light of the newness of health-care data warehouses. "There are core requirements that you think you have to have, and when you start digging in, it's more difficult than you thought. When I arrived, there were no Internet connections, no Unix network, no machinery. We got it in January and had the first connections to the database in June." The development could have gone a lot worse, he adds. "There are organizations many times larger than Market Insights that made the same platforms decisions we did, and they were six months behind us."

Despite the conflicting expectations and lack of progress, Louie and Milano wanted Bandt to stay on. The modest salary they were paying him was affordable, and they wanted the project done. But the three were unable to agree on the equity stake Bandt would receive. In October he left to join Healtheon Corp., in Santa Clara, Calif. "It was probably a lack of experience all around in understanding how long [it would take] and how complicated some of these things are," Bandt says now. In hindsight, he adds, "I don't know if I would have made the same technology decisions, but I was the only one who could make the decisions. There really wasn't anyone else around for me to talk with about these things."

By the end of 1998, more than a year after the project's launch, nothing had worked out the way it was supposed to. Bandt was gone. The data warehouse was far from finished. Inpatient and outpatient databases for all states had been loaded in, along with the financials for all U.S. hospitals. But outside the four walls of Market Insights, users still had minimal access to the warehouse. The ability to do sophisticated analyses that Louie and Milano had envisioned was still a pipe dream. And stuffed with data, the warehouse performed sluggishly.

Saddled with expensive equipment and no service to sell, Louie and Milano had to regroup. All of 1998's profits had been sunk into the warehouse. After Bandt left, the founders accessed a line of credit for an undisclosed sum. They used the additional funds to hire programmers. By year's end, labor costs for the project tipped the payroll at $250,000. Even worse, deploying programmers to the warehouse project left existing, more salable products lagging and unfinished. A grim Milano found himself explaining the project's delays to customers to whom he had proudly talked it up only months earlier.

Louie and Milano have taken several steps to redirect Market Insights and craft a strategy to stabilize the company and complete the warehouse project. A new chief operating officer, vice-president of sales, and four sales reps came on board early this past summer. The company also dipped into its incoming revenues to hire another two full-time programmers to fine-tune the software and begin speeding up the data warehouse so it has more appeal for users. That move freed up the company's longtime in-house developer to work on new software products.

Bit by bit, Market Insights is beginning to wring sales from the warehouse effort--though not necessarily in the way its founders had hoped. Early proponents of data access waxed on about its egalitarian nature. Throw open the doors to industry data collections, they predicted, and there's no limit to what will be discovered. The reality has been much more pedestrian. "People don't want all the information that's out there," says Louie. "They want a middle person like Market Insights to make sense of it for them." He still feels the company's future lies in one centralized Internet application that allows access and analysis of all Market Insights' data, "but it's further away than we anticipated. I mean, think about it: our clients are CFOs. They're high up in the organization, and they don't have a lot of time on their hands. Not all of them have Internet access, and those that do, don't want to do analysis."

Louie calculates that Market Insights would need 100 users paying an average subscription fee of between $5,000 and $10,000 each in order for the project to begin to break even. "I don't see 100 users logging on at this point," he admits. So Market Insights is using the data warehouse to branch out from its core hospital and provider clients. Last spring the company launched National Hospital Almanac, a $1,200-a-year subscription service that's designed to appeal to anyone related to health care: HMOs, PPOs, medical-device manufacturers, even investment bankers. It offers unlimited Web access to portions of the data warehouse. But the Almanac is a Java application--Market Insights still has no Information Advantage-based products or services up and running--and allows little more than a static look at the data.

Had he to do it over, Milano would have planned in more detail--and put everything in writing. Along with Bandt and Louie, he agrees that the decision to use Unix--a complicated operating system that no one had much experience with--made everything tougher. But he still believes fervently in Market Insights' data warehouse as the future of the business. And maybe that's the lesson here. Technology is immensely powerful--and immensely seductive. Handling a major new technology project in-house can seem sensible, expedient--and perfectly feasible. It's all too easy to ignore the inevitable problems and risks that face even an experienced company.

"Our decision, which we don't regret, was that although we could have stayed a nice mom-and-pop store where we each made a good living, we wanted to become a supermarket," says Louie. "We took the risk, and we can realize the reward. It's just not what we had hoped for."

Deborah Asbrand is a freelance writer based in Boston who covers technology and business.