[CSGP – CoStar Group; REIS – Reis Inc.] Tale of Two Data Providers

Before CoStar came onto the scene in 1987, getting clean and current data on rental rates, vacancies, absorption, and precedent comps needed to transact in commercial real estate was a vexing, ad hoc process.  In the early/mid-90s, juniors staffed across various financial institutions devoted significant chunks of their workweeks collecting and scrubbing this information.  Or they otherwise relied on multiple third party collators who each focused on a limited number of markets and applied their own idiosyncratic methods with varying degrees of breadth and accuracy, yielding a byzantine patchwork of non-comparable data, updated only every 90 days or so, across markets.

CoStar has painstakingly built its database – hundreds of fields like site, zoning, tax assessment, deed, sales comps and ownership information and photographs and floor plans updated thousands of times per day across 5mn properties – through both acquisition and the brute dedication of full-time researchers who conduct millions of interviews with industry participants, physically inspect hundreds of thousands of properties, even hop on board camera and sensor laden aircraft to procure videos and photos and scout new developments, while a gamut of automated data quality tests check for errors.

Thirty years of this, plus ~100 proprietary datasets picked up through acquisition, has generated the largest proprietary commercial real estate database in the world.  CoStar’s data is woven in into the workflows and in-house datasets of a fragmented landscape of brokers, owners, REITs, property managers, investors, governments, and banks, among others [no customer is > 5% of revenue.  At the time CoStar went public in 1998, brokers comprised 80% of its revenue; today, they are less than 30%] that handled over 80% of all commercial real estate transactions in 2016, and is so ubiquitously referenced as to have drawn accusations of abusive monopoly power from competitors for years (and was even issued a consent order by the FTC in 2012 in connection to its merger with LoopNet), often in response to the company’s spirited legal enforcement of its IP (CSGP has filed at least 32 lawsuits, mostly related to copyright infringement, since 1999).  As you might expect, CoStar’s information and analytics customers are sticky, with retention rates running in the mid-90s.  And despite the highly cyclical end market it serves, subscription renewals are resilient to downturns, with revenues declining less than 2% in 2009 and growing in 2001-2002.

If you rewind to 1998, you’ll see that REIS, a publicly traded comp to CoStar that I’ll get to later in this post, was over 30% larger than CoStar by revenue thanks to its 7 year head start.  But over the subsequent 20 years, both companies pursued divergent growth strategies – REIS expanding its datasets steadily and organically; CoStar aggressively acquiring its way into new geographies, real estate verticals, and product markets – that would translate to devastatingly discrepant value creation profiles and leave REIS languishing woefully behind CoStar’s lambent surge.

Buying local databases, it turns out, made a whole lot of sense because by merging acquired databases with its own, CoStar could upsell new coverage to its existing customers and cross-sell existing data and analytics to customers of the acquired one, with an increasingly well-populated central dataset promoting customer retention all-around, especially as commercial real estate brokers themselves consolidated into larger regional and national players in need of granular detail and broad coverage.  Building data assets in new markets from scratch, on the other hand, consumes enormous time and resources, and given how sticky customers can be, getting there first with a more complete solution, matters.

While investors, especially value types, are right more often than not to “tsk-tsk” acquisition-driven growth, REIS did itself an irredeemable disservice by sticking to its ascetic home grown discipline.  Some commercial real estate professionals will claim that the two companies’ services aren’t even all that comparable today, with REIS offering higher-level trend data for executive-types and research/appraisal departments in banks and brokerages, and CoStar providing the integral, digitally delivered nitty gritty for front line sales brokers deeply immersed in commercial real estate every working hour.  A professor in business school once told me that the value of a business can be gauged by the number of tears shed in a world deprived of its services.  It seems clear which company’s services, if removed, would provoke more wailing.

For most of its early life, the preponderance of CoStar’s acquisition efforts were were devoted to database assets.  That changed in April 2012, when the company acquired LoopNet [paying 9x revenue and ~30x EBITDA], the first of several ostensibly non-adjacent marketplace acquisitions that would dramatically reshape its corporate profile.  [LoopNet is a web-based marketplace for commercial real estate properties.  For monthly subscription fees, landlords and commercial real estate brokers can list properties for sale and lease to an audience of prospective buyers and tenants…like Zillow, except for office, industrial, and retail properties.].  CoStar’s listings marketplaces now account for nearly half of total revenue, up from nothing 5-6 years ago, and are growing substantially faster in a less penetrated market than the core Information and Analytics business.

Besides the cost synergies from leveraging the combined technology infrastructure and other backend expenses – which brought $20mn of intermediate cost synergies vs. LoopNet’s then $28mn of EBITDA – realized in half their targeted time, the acquisition promised even bigger upside from putative cross-selling opportunities between both companies’ non-overlapping subscriber bases [CoStar’s customers tilted corporate and its properties were larger buildings in urban areas while LoopNet sold more to individuals, its listings more main street heavy], a belief bolstered by surveys purporting that the 10% of clients who subscribed to both LoopNet and CoStar pre-merger claimed a different set of benefits from each and so would continue subscribing to both post-merger.

Prior to acquiring LoopNet, CoStar actually had its own commercial marketing service – CoStar Showcase, launched in 2008 – that I don’t think was very well adopted.  LoopNet offered those CoStar subscribers a fresh audience of 2mn monthly uniques (now 5mn), peanuts perhaps in consumer space but significant in the world of chunky, big-dollar commercial real estate.  More meaningfully, LoopNet’s 100k repeat search members, a sophisticated subset of LoopNet’s users with more intensive needs, were targets for CoStar’s information products: LoopNet delivered a broad, marketable audience; CoStar brought superior data and analytics.

Furthermore, there were opportunities to directly improve the standalone site.  LoopNet’s previous management badly neglected its higher commission / lower churn Premium Lister service, which experienced revenue declines from 2008 to 2011 while competitor marketplace services grew.  CoStar’s management thought it could push listing penetration (its share was ~10% in 2012) by boosting marketing spend and re-orienting sales incentives to its ~80k Premium Listing subscribers; improve churn by selling to companies vs. individuals and lengthening contract terms; and pull more value by raising prices for early vintage clients (many of whom at the time were still paying less than half the current list price) and by replacing LoopNet’s subpar information products with their own.  And they did a fine job in the 2 years post acquisition, reducing Premium sub churn, boosting average revenue per sub by 60%, paid listings by 55%, monthly uniques by 37%, and nearly doubling LN’s organic revenue growth rate from 10% at the time of acquisition to ~20%.

It’s probably fair to say that revenue synergies in the first few years fell somewhat short of expectations.  At 2016 year-end, only 18% of CoStar’s clients were also LoopNet Premium Listers and 26% of LoopNet’s paying customers were also CoStar subscribers.  At least part of the issue was that up until just a few months ago, the LoopNet and CoStar databases, each with their decades worth of data, still operated separately for the most part…so the valuable incremental content that CoStar brought to LoopNet – information on fully leased buildings, tenants, sales and lease comps, as well as a slew of analytics – was not plainly obvious on the LoopNet portal, confounding upsell efforts.

But even so, by the end of 2016, the company realized an incremental $80mn of cross-sold CoStar subscription sales.  Conservatively assuming those revs came on board at 25% EBITDA margins and including the $20mn in realized cost synergies, that’s an additional $40mn of EBITDA on top of the $28mn of LoopNet’s standalone EBITDA at the time of acquisition…and this doesn’t even include the significant aforementioned self-help improvements.  Now the $880mn purchase price doesn’t look so crazy.

And that’s just for starters.  Now that the databases have finally been integrated and low-value competitive products from LoopNet have been purged, the company can more readily showcase CoStar content to LoopNet users [LoopNet users who aren’t CoStar subs will see a “locked pin” next to CoStar content in the search results] and more easily penetrate the LoopNet base with higher-priced subscriptions.  And apparently cross-sales are going gangbusters so far, with close to $3mn of net new cross-sold bookings in the first two weeks since integration (vs. $80mn in cross-sales during the prior 4 years) and the average cross-sold LoopNet client paying nearly 6x as much as before ($80/month to $470/month) [note: this is the monthly rate per client, not subscriber.  The distinction matters because CoStar is sold to enterprises under license agreements with multiple users per license, while LoopNet was sold to individuals].

But perhaps one of the more interesting but less oft-mentioned angles to this acquisition – and other lead gen services that the company has since acquired – is the degree to which new commercial properties can be cost-effectively added to CoStar’s dataset.  Just as the insurance industry feeds its claims data to Verisk for free (which data Verisk then sells back to the industry at a huge mark-up), brokers who list their properties on LoopNet are proactively feeding data on hundreds of thousands of properties to CoStar’s database, which work otherwise would have been borne at significant cost by in-house researchers.