The Rent Roll Is the New Balance Sheet

At 8:47 on a Tuesday morning in September 2019, the reception desk at the New York headquarters of a midtown brokerage got a call it had received three times that month. The caller wasn't a landlord or a tenant. It was a data broker, offering to sell the firm's own leasing activity — floor by floor, rent by rent — back to it, cleaned and priced.

That call, and a thousand like it, explains why the most valuable thing in commercial real estate is no longer the building. It's the information about the building.

For a century, the rent roll was a private document — a landlord's ledger, kept in a drawer. Today it is a product. The company that figured this out first, CoStar Group, turned the industry's private ledgers into a business that posted nearly $2.7 billion in revenue in 2024 and carries a market value in the tens of billions — roughly the worth of a midsize REIT, built entirely out of other people's buildings' numbers.

The mechanism is elegantly old-fashioned. Founded in 1987 by Andy Florance with a mortgage-default database, CoStar spent four decades doing the most boring, most expensive thing in the business: verification. Thousands of field researchers and analysts call brokers, photograph buildings, confirm leases, and reconcile the conflicting versions of reality that landlords, brokers, and tenants each tell. By the time CoStar's machine is done, the rent a tenant paid in Dallas and the vacancy rate in downtown San Francisco exist in the same schema, updated daily.

That daily verified record is the moat, and it monetizes like one. CoStar's customers — brokerages, lenders, appraisers, institutional owners — don't buy data the way they buy software. They rent the truth, annually, at prices that only move in one direction. The subscription renewal rate in CoStar's core information-services business has historically run in the high 80s to low 90s percent, a number any enterprise software CEO would envy. When the same customer pays a five-figure annual subscription for a decade, the data isn't a product anymore. It's infrastructure.

The acquisitions tell the story of a company buying not revenue but redundancy. LoopNet, bought in 2012 for roughly $860 million, gave CoStar the marketplace where the listings live. Apartments.com, acquired in 2014 for $585 million, gave it the renter. STR, the hotel data benchmark, cost $450 million in 2019. Ten-X, the auction platform, was $190 million in 2020. Each deal folded another stream of the industry's private numbers into the same vault. Competitors didn't lose because they had worse software. They lost because the industry had already standardized on CoStar's numbers, and changing the standard is far more expensive than paying the subscription.

Rent data is particularly sticky because it feeds everything downstream. A lender's debt-service coverage calculation, a REIT's net asset value, a broker's opinion of value, an appraiser's comparable-sales adjustment — all of them quietly assume access to a verified lease database. When Moody's paid approximately $278 million for REIS in 2018, the oldest commercial-property research firm in the country, it wasn't buying software either. It was buying the feed its credit models depend on.

There is a counterargument, and it has a name: CompStak. The crowdsourced lease-comp platform built a dataset of actual executed leases by giving brokers the comp data in exchange for contributing their own. Its success proves the moat leaks — brokers have always traded information, and a clever exchange model can assemble a parallel vault at a fraction of CoStar's cost. Public records, SEC filings, and now AI extraction of offering memoranda are all solvent dripping on the same wall. The raw facts of commercial real estate were never proprietary in any absolute sense; the proprietary part was the labor of knowing them.

But "leaky" is not the same as "broken." What CoStar actually sells isn't facts — it's confidence. A portfolio manager marking $10 billion of assets to market doesn't need a number. She needs a number nobody will challenge. CoStar's decades of verification history are what makes its number unchallengeable, and that confidence is what renews at 90 percent. CompStak nibbles at the edges; it hasn't replaced the standard.

The deeper lesson of the story is about the nature of the asset. A rent roll is not like software, which decays when the architecture ages. It's like a gold mine that gets richer as it gets dug: every new lease record makes the historical dataset more valuable, because forecasting requires history. The dataset's marginal cost falls while its marginal value rises. That's why CoStar can spend more than a billion dollars building Homes.com and still be rational — it's buying the future residential rent rolls before anyone else digs them.

Consider what that implies for a name like this one. The next rent-data franchise — built on AI extraction, crowdsourcing, sensor data from buildings themselves — will need distribution, and distribution starts with trust. In a market where buyers rent the truth, the address matters: the exact, category-defining name is the cheapest credibility the business can buy. The vaults get deeper every year. The doors to them are getting narrower.