Andrew Florance looks back on CoStar Group’s origins — his Princeton dorm room circa 1986 — and jokingly asks, "What was I thinking?" Not many college juniors have the foresight to start a legitimate business, much less a company that eventually grows into commercial real estate’s dominant information source. As an aside, Florance, CoStar’s CEO, admits that he did actually have some fun in college.

Now, Florance is having an equally good time fashioning CoStar into one of the most successful online ventures around, regardless of industry sector. The company is poised for profitability — yes, real profits — by the end of 2001. Its research force, now 600 strong, maintains perhaps the most comprehensive and up-to-date source for property information available.

In between trips to Scotland, where he’s getting married this fall, and San Francisco, Florance took a few minutes to discuss CoStar and e-real estate with RETech.

NREI: Tell us about CoStar Connect and how it changes and improves the services CoStar offers.

Andrew Florance: We have a number of products we’ve developed over the years. Pretty early on, you can get a sense as to which products are going to be extraordinarily popular, and this is one of them. It’s a simple concept that’s compelling to a brokerage firm. Version 1 of CoStar Connect is just the beginning of a whole new philosophy of product that basically enhances our brokerage clients’ ability to communicate with all their different constituencies professionally, effectively and cost efficiently.

We already have planned Version 2, Version 3 and Version 4 of CoStar Connect, integrating with various products, but we’ll focus on Version 1. If you pull up any given brokerage firm’s Web site in the U.S., you’ll find that less than 30% of the listings they represent are accurately reflected. As well, the software interface is generally pretty rudimentary, and generally there won’t be images of the properties. Sometimes there will be, sometimes there won’t, and it will be of different quality.

There was a first generation of software systems that came out in 1995 and 1996, like LoopLink. Market-to-Market came out in ’98, and there were a couple other variations of that, such as Comro and others.

What they offered as a value proposition was a simple software interface where you could key your own listings into your own Web site. That was good stuff for 1995 and 1996, and it was affordable. For a couple hundred dollars a month, brokerage firms could get their listings up on their site and not have to hire a software developer to do it.

But people discovered that kind of solution doesn’t work because, if you’re even a mid-size brokerage firm, your brokers are basically independent contractors. They come and go as they please. They have all kinds of things to do.

If you’ve got a broker who brings in $1 million in commissions a year, trying to get that broker to take a camera out and photograph all the listings, scan the photo, upload it to the Web and key in all the data and update it every 30 days is worse than herding cats. It’s not going to happen, and the concept of a manager firing that broker because he didn’t take that camera out and take the pictures is absurd. That broker can do anything he wants.

That’s where the huge flaw exists in the first-generation systems. You can have great fun with it. You go into some big brokerage firm and they say they have their listings on the Web and you pull them up and they’re not there. They’re all wrong and out of date. It’s a problem.

What we specialize in as a company is trying to build a relationship with that million-dollar producer, find out where they are, and get the information from them. We go out in the field and take the photo for that million-dollar producer. If we can’t get that million-dollar producer, we hunt down his assistant. We hunt down anyone who knows anything about that listing, and we get the information in there on a timely basis and think about that all day. That’s what we’re paid to do.

CoStar Connect brings out a second generation of software that is much more user-friendly for marketing your own listings on your own site. We’re giving our customers the assurance that all of their inventory is up there, that it is proactively managed, that high-quality images are up there and that the entire inventory is up there. It’s really cutting-edge technology.

When the first-generation software systems came out, they weren’t sure if they were providing an information service similar to a CoStar Property, one of our basic information services, or if they were a marketing service because it was so early in the industry’s development. They weren’t sure who their audience was. They weren’t sure who they were trying to reach.

Now, it’s become clear who we’re trying to reach when you’re marketing a brokerage firm’s listings. You’re trying to reach the CEO who’s driving down the street and sees the leasing sign that has a Web address on it, and he wants to browse the property. That’s a real novice, and the software interface has to be something that looks more like a J. Crew site or a Sharper Image site. It shouldn’t look like such a technical terminal, which a lot of the first-generation systems looked like.

We went for a really consumer-friendly look and feel — high-quality content, high-quality imagery — that allows the brokerage firm to look much more professional in their prospects’ eyes. That’s what we think we have with Connect.

We’ve been signing up one Web site every other day since we released it in April.

The next generation is a whole series of communication tools that take you through the process via the broker’s Web site. Suppose a lead comes to me off my Web site, or even doesn’t come to me off my Web site. If I have a tenant that’s looking for space, I’m going to use CoStar Property, which is my only solution for having a complete inventory in a given market. I’m going to search for all the buildings that meet that tenant’s requirements. I’m going to do an awful lot of communication with that tenant, the tenant’s CFO, the tenant’s facilities manager, possibly the tenant’s CEO in another city. There’s going to be a lot of going back and forth, a lot of feedback from the tenant.

We’re building the ability to publish groupings of buildings or surveys of buildings and schedule building tours and report to the tenant through the Web site. Through Connect, the tenant can access the information in real time and see what group of buildings the broker has selected for the tenant, see all the feedback that’s entered about the buildings and just really open up the process more to the tenant through the broker’s Web site. They’ll basically be able to manage the transaction through the broker’s Web site. Developing those kinds of software tools is a multi-million dollar effort. That’s what we perceive our role to be — to make it easier to publish your content, manage all your imagery, manage the basic data, harass all the brokers to get the information, and spend enough money using economies of scale to actually produce the kinds of cutting-edge tools that produce efficiencies for the industry.

We’re pretty excited about Connect. It’s received a straight A+ response.

NREI: At this point, can you say more about future versions?

Florance: The future versions will allow the broker, for a pretty small fee, to market all their listings to tenants with high-quality photographs. We manage all the data for them. Future versions allow them to report to their owner-clients or their broker-clients, and share information from all of our clients, through the Connect interface with their clients. A client can come to a broker’s site, enter the password and see when they’ve been scheduled to tour different buildings. They can actually see different request for proposals (RFPs) that have gone out, see what offers have come back from different buildings and see discounted cash-flow analysis off those different offers.

It’s all about investing the money to make our clients look more professional. We’ll be background people. The whole name CoStar reflects the fact that we’re a background player. It’s a way of sharing and opening up all of our information for the broker to be able to share it with their clients on the Web.

In a given lease transaction from, when you decide to try and find 5,000 sq. ft. in a particular city to when you actually see the tenant walk in the door, you have at least 1,000 lines of communication that are going to occur. Historically, a lot of that communication been via paper or fax, e-mail, hard copy or Power Point presentations. [Connect] will cut that down dramatically and allow people to do a lot more of that on the Web.

I need to stress that this does not change the role of the broker. CoStar Group is one of the few companies that actually is completely broker-centric and is geared toward the broker. We don’t view this as being online leasing. These are brokerage tools.

NREI: CoStar experienced an acquisition flurry a few years ago, and now we see another flurry of consolidation and M&A activity, as well as a few companies falling by the wayside. What’s the next stage?

Florance: Bank of America said there were about 450 e-real estate companies in our space at the end of ’99, and they now say there are less than 150. You’re talking about amazing attrition.

We acquired 15 companies over the past 10 years. Everyone that got in trouble in the past 18 months sent their business plan our way. I’ve literally seen 30-plus business plans from companies that are falling by the wayside. I think we’re ending one period, and we’re not quite out of it yet. The next phase hasn’t arrived.

I’ll break the cycle into three phases. There was a period when the first wave of consolidation went on, and we acquired Jamison, Lease Trend, Chicago Resource, Comps, where we were out there acquiring companies that had revenue. We were buying those companies for a multiple of 2.5- to three-times revenue. So all the companies we bought gave us data, gave us clients and gave us revenue.

We never purchased anything that didn’t have revenues or clients. We never paid some outrageous number for a company. In the past year, the 30 or 40 companies that have presented their business plans to us have given us the opportunity to acquire companies that have no material revenues, no constituency of clients who need their products and services. These opportunities often required us to pay ultra-high multiples, or ultra-high prices, for companies that are losing tons of money. If you’re doing M&A right now, you’re supposed to pay still-inflated prices for the chance to lose money. Maybe I’m stupid, but I don’t get it.

We actually looked at an acquisition recently where the venture money behind it was almost willing to pay us to acquire the company. It’s known that RealtyIQ, which was a company that had well north of $70 million invested in it, recently sold for well less than $4 million. Even so, you’re buying a company that lost $50 million last year.

With PropertyFirst and LoopNet’s merger, that’s a logical deal, but the high-profile venture capitalist behind PropertyFirst knew that there wasn’t any significant revenue in PropertyFirst. There were huge burn rates. Rather than shut it down, if you roll it into another company you don’t have to take the write-off. You don’t have to admit that it failed.

From LoopNet’s perspective, it’s running out of cash and still has a huge negative burn rate. PropertyFirst still has cash. It keeps your doors open longer.

There’s a lot of M&A work going on right now, but it’s basically brilliant press releases for disguised Chapter 7 filings. It’s like a fancy bankruptcy notice.

The companies you want to see are the companies that have solid business plans, where they’ve proven that they can make money and be profitable or that they’re close to being profitable and have a track record of making money. There are only a small handful of companies like that out there.

If you take Bank of America’s study — the 450 companies now reduced to 150 — I think we’re probably going down to 30 or 40 viable companies in this space. That will probably take another 18 months before we weed out the rest of the ones that are going to die anyway. Then you’ll begin to see real M&A begin again. Again, it will be strategic and sensible M&A activity.

RETech: What are the advantages and disadvantages of mining and owning your own data vs. the self-populating systems out there?

Florance: I don’t agree with the premise of the question, but, having said that, let me wander around that subject.

First, a synonym for a self-populating system is a perpetual motion machine, and there’s no such thing. Another synonym for a self-populating system would be passive collection.

I don’t quite understand what someone means by the advantages and disadvantages of owning the data. It’s something you use in a press release. You use that if you are a company that has no revenues and wants to accomplish something that’s attractive to a potential audience. It doesn’t have a lot of substance behind it.

Let’s look at the issue of owning data. Does CoStar Group own data, and is it possible to create self-populating systems where the people that populate own the data? Is CoStar a different kind of model where CoStar goes out and does something called research and then owns the data? You often hear from a prospective client, they’ll say, "Gosh, you take the data from me, and then you sell it back to me."

Well, no, we’re not going to charge you for the data you gave us. We’ll give you that back for free. We’re going to give you a half-percentage point discount on your service to reflect what you gave us. We’re just going to charge you for the 99.5% of the data that we spent $90 million last year to collect from other people.

If you look at our software system in any given city, only a tiny fraction of it comes from any one firm. We’ve hired a photographer to go out and take a photo of the building. We’ve hired an airplane to fly over and photograph the building. We sent someone to the courthouse to look for deeds to see what ownership transfers and mortgages are in there. We’ve phoned all the tenants in the building and asked them about their leases, and we’ve usually tracked them for years. We’ve paid hundreds of thousands of dollars to different groups to access public records and sort through them. We have talked to the owner and the property manager. We drive 2 million miles a year, driving by the building to learn about it. We talk to more than 1.5 million different people a year. There’s no one source or person giving us any more than .05% of our data.

Yes, once we spend $90 million on collecting content and photographs and the like, we provide a valuable service that people pay us money for — to do all this hard work. There’s no silver bullet or perpetual motion machine that makes this all magically pop up.

On the issue of the self-populating system, when you talk about the advantages vs. the disadvantages, that implies that we think we own all the data or that we’re not a self-populating system. We are, in fact, a self-populating system. Users can hit a key in our different software products and electronically transmit data to us as they wish. You can go to our Web site and add a listing just like you add a listing on any other self-populating system. You can send us your broker mailer. You can voluntarily give us information electronically, by fax, e-mail, whatever form you want to use.

We’re a self-populating system, but we don’t stop there. We take the self-populating information, and then we spend $90 million on top of that to pursue the data that people didn’t volunteer to us. You can tell us about the listing. You can electronically tell us about the listing. You can fax us about the listing. But if you don’t, we’ll also be driving past the building and calling you. If you forget to call us, we’ll call you.

For example, from an information system, users need to be able to know who tenants are in the building. No building owner in his right mind is going to self-populate the tenants in the building. No owner in his right mind is going to say, "These are my clients. This is what they’re paying." No owner that just sold his building in a distress sale is going to self-populate that distress sale.

We are a self-populating system, but that’s probably about 30% of our business. The reason we have a viable, real business model is that we do the other 70% of the work that never will self-populate, and it gives you the whole picture.

This debate over the self-populating system, which began six years ago, shouldn’t be continuing. LoopNet and CoStar Group both began their rollouts at about the same time about six years ago. One was based on proactively collecting valuable data through a combination of active and passive research, and the other system was about just collecting data through passive research. Six years later, one of the companies has gone public and has $70 million in revenues, and the other company has a couple million in revenues and is still espousing that this self-populating system is the way to go. But they’re not going anywhere.

RETech: What are you doing to continually improve the accuracy and timeliness of your data?

Florance: We are not satisfied with where we are. We need to be to the point where 85% of the work involves us going out and hunting for the data. We’ve awakened a huge industry to a new way of doing business, where the data is efficient.

They know what they want now. They want more timely information, more comprehensive information, and the standards are ever increasing. We as a company are taking that challenge. Our goal for this year is to improve our data quality by 20% by the end of the year and try to do that every year on out. At the same time, another primary mission for the company is to keep our prices flat while we continue to improve data quality. I would say the top of my agenda as CEO of CoStar Group is to do what we know we can do, which is improve data quality another 20% to 30% this year and do that without increasing the costs to our clients $1.

There are a number of ways we’re doing that. To put things in perspective, two years ago we had 120 researchers. As you can appreciate, any time you go from 120 researchers to 600 in a two-year period, you’ve got to build your best and brightest managers, invest in your people, train them and bring them along. They need time to develop relationships with the brokers out there across the United States.

We’ve long known that you’re most vulnerable as far as data quality in the first 24 months, and then data quality continually improves because you’re no longer trying to start up in the market. You’re actually building it up. We’re also investing a lot more in our researchers’ training and incentive programs.

We’ve developed some amazing software called benchmarkware. We’re actually quantifying 1 million different data points of quality in our database at any given minute. We’re whittling those quality levels down to the specific researcher, the team, the managers, the overall division. We can pull down a real-time report on where we are regarding the quality level on a scale of 1 to 100 by researcher, by city, by market, by product. That allows us to target where we need to improve, where we need to add additional staff and where we need to shuffle staff. You can actually see an individual error rate for a particular field independently developed at any given time and how we’re improving that.

Another big part of data quality is communicating with your clients and training your clients. We have a quality-control team of about a dozen people that independently audit data quality. Our No. 1 cause of quality issues is client training.

We added 10,000 to 15,000 new users and brokers, last year on a complex, robust software system. Training is a big priority so people know how to use it. We’ve actually brought in [former Comps president] Michael Arbe out of the Comps division to run our overall customer service. We have a much different focus on improving our customer service. We are much more proactive about being out in the field training our clients.

Continuous quality improvement is our No. 1 focus. We think that we’ve been fortunate as a company. The industry has given us the opportunity to be the leading player in the market, and 35,000 brokers are under contract, purchasing our products and information services. We take that responsibility and our role seriously and work our hearts out to deliver to the brokers the quality they deserve for what they’re paying in fees.

RETech: What’s the timeframe for profitability, and what are the keys to making money with what is, at its core, an online venture? Has the Nasdaq slump limited any of your plans?

Florance: The timeframe to profitability is fairly straightforward. Our stock price went up about 30% after we released our [first-quarter] earnings. One reason people are excited is that our losses decreased 31% in one quarter.

Four quarters ago, we were losing 65 cents per share. Three quarters ago, we were losing 55 cents per share. Two quarters ago, we were losing 43 cents per share. This past quarter, we were losing 29 cents per share. You have a predictable pattern. Each quarter we’re adding about 10 cents per share of improvement. Out of the 50-plus markets we’re in, 40 of them are already profitable. It takes time. At the end of this year, we will be cash-flow positive. Also, the company has ample cash reserves, more than $43 million still in the bank. With narrowing losses, we have a solid cushion in cash balances.

When we go into a city, we invest a ton of money upfront to open an office, to train sales people, to initially drive by all the buildings. You have no subscribers at that point. Then you build up a handful of subscribers each month. After about 18 months, you have enough subscribers in place to cover the fixed costs of providing the service. Really, we’ve had losses over the past two years because we opened up in so many markets at once. Now that we’ve actually been out there in these markets for more than a year, now we’re moving back toward profitability.

RETech: It also goes back to, if you acquire companies that are already profitable, that obviously makes it easier.

Florance: We acquired Comps in February of last year, and it was losing about $2.5 million a month. We looked at that and determined that we could peel away a lot of the dot-com initiatives, and there’s profitability inside that. We did, and in one year Comps went from losing 2.5 million a month to the point where it is now profitable.

The key to profitability is really simple. You have to test your concept first on a small scale, make sure it works, make sure it makes money and then grow the concept. If you take a theory and you just try to raise a tremendous amount of money and try to develop a complex national operation and then try to figure out if your theory was right or wrong on top of that big, complex national operation, you get in big trouble.

The banner for that kind of problem is Webvan. It never tried to do grocery delivery in one city to determine if it could make money. Instead, Webvan raised $1 billion and tried to implement it across the country at once. The company discovered that it didn’t make money. I think there’s a lot of that out there. The key is to actually prove to your investors and to yourself that your product makes money.

We were profitable in Washington, D.C., before we went to New York. We were profitable nationally before we went public. And companies like Comps were profitable. Jamison had been profitable. You don’t rely too much on theory. You say, "Prove it to me," with your business model.

The Nasdaq slump has affected us in both positive and negative ways. On the negative side, we, like most everyone else, hate to see our stock sliding down with all the other technology issues. That’s no fun. Not being able to raise capital isn’t really an issue because we have plenty of capital.

It changed the ethic out there in the business community. Before, the big question was how aggressively you were seizing new territory and how aggressively you were investing to capture future advantage? Now, the question is what does your profitability look like today? If the Nasdaq had kept rising to 6,000 and 7,000, I bet we’d be opening offices throughout Europe. Because it didn’t, we’re focusing on quality improvements and on profitability and efficiency in our business model here in the United States.

Probably the biggest advantage of the Nasdaq slump is that we have a number of competitors out there, and it basically required them to prove their business plans, too. A lot of them fell by the wayside when they had to prove their business plans.

RETech: What advice would you give to a budding tech entrepreneur?

Florance: It has to be a calling. If you’re getting into it because you think you’re going to make a quick buck, it’s a bad place to be because you have to have real commitment to what you’re doing. It can’t be just about the money. It has to be about the impact you have on the industry, or about how you change it for the better. The reality is, 99% of the people who get into this business are going to get burned. You have a better chance of being hit by lightning than being an e-real estate tech success.

Having said that, you can’t lie to yourself, you can’t lie to other people. You have to prove your business model, and you don’t want to scale the business model until you try it on a small scale first. You have to prove that it works with real clients with real money buying your products and services. Nine out of 10 entrepreneurs over the past year have been going out there and creating businesses that only work on paper. They don’t work in the real world. A lot of them are still doing that.