CoStar Group, Inc. (NASDAQ: CSGP) — The Adjusted Breakeven Was Real; the Growth Proof Still Is Not

Research date: September 2, 2026. Market data through the September 1 close. Dollar amounts are U.S. dollars unless stated otherwise.

⚡ Claude’s Take

This block is the author’s independent opinion and general information only—not investment advice. The analytical sections that follow take no position and carry no price target; the single opinion in this report is confined to this block.

Verdict: HOLD/WATCH at $31.98; accumulate only on a retreat into roughly the $26–$29 range; conviction MEDIUM-LOW. This is a downgrade in posture from the July 3 view, which favored accumulating below about $32, even though one important piece of that thesis worked. Residential Real Estate produced $12 million of adjusted EBITDA in Q2, versus a $76 million loss a year earlier, and management kept its $780–$820 million full-year adjusted-EBITDA outlook. The commercial engine remained excellent: $481 million of revenue, a 36% adjusted-EBITDA margin, 93% quarterly CoStar renewal and 19% subscriber growth. Those are not the figures of a broken information franchise. They are evidence that the core still has the scale economies and customer captivity that made CoStar unusual in the first place. CoStar Q2 release

But the price of certainty rose just as the operating story became less bad. Actual Residential EBITDA was still a $5 million loss; Homes.com subscribers increased only modestly from 35,175 in Q1 to just over 36,000 in Q2; bookings were merely similar sequentially after a May price increase; and consolidated net-new bookings fell 26% year over year. Management cut revenue guidance to $3.715–$3.755 billion while preserving profit guidance through roughly $100 million of planned expense reductions, including a sharp contraction in the Homes.com sales force. That is a credible profitability turn, but it is not yet proof of scalable customer acquisition or durable marketplace economics. Q2 10-Q Q2 earnings event

The second change is capital allocation. On August 21 CoStar paid $800 million cash for Zonda, a useful new-home data and workflow asset with about $170 million of 2025 revenue and a 23% adjusted-EBITDA margin. The price—about 4.7 times revenue and 20.5 times adjusted EBITDA before synergies—was not reckless for a strategic information asset, but it consumed most of the $1.266 billion of June cash after CoStar had already spent $589 million repurchasing shares in the first half. A static post-close balance-sheet estimate moves CoStar from modest net cash to roughly $0.5 billion of net debt before subsequent cash generation and deal costs. Residential is therefore no longer a “free option” sitting beside a cash fortress. Shareholders now own a serial acquirer simultaneously integrating Domain, Matterport and Zonda while funding Homes.com. Zonda closing 8-K Zonda release

At $31.98 and 405.2 million latest reported shares, the equity is worth about $13.0 billion. On a static post-Zonda basis, enterprise value is approximately $13.5 billion, or about 3.6 times pro-forma revenue and 16 times the combination of CoStar’s 2026 adjusted-EBITDA midpoint and Zonda’s 2025 adjusted EBITDA. That is dramatically below CoStar’s former double-digit sales multiples, but those pandemic-era valuations are the wrong anchor. On actual trailing EBITDA and owner cash flow after stock compensation, the shares are not plainly cheap. The market is already underwriting a large margin recovery.

My entry discipline therefore tightens. The attractive setup is still visible: a wide-moat commercial-data franchise, a strong Apartments.com marketplace, a falling capex burden, fewer shares, and meaningful operating leverage if residential revenue survives the expense reset. The problem is that today’s price asks investors to capitalize the adjusted recovery before Homes.com has disclosed customer-acquisition cost, cohort retention, depth-product adoption or standalone profitability. The evidence that would change this view upward is two more quarters of positive actual Residential EBITDA, reaccelerating net bookings, continued Homes.com run-rate growth with churn at or below 2.5% monthly, and buybacks funded by free cash flow rather than cash depletion. The evidence that would change it downward is renewed Residential losses, weakening CoStar/Apartments renewal or price, continued acquisitions before integration returns are visible, or search platforms taking a rising share of portal economics. The commercial franchise deserves patience; the capital allocation record demands a margin of safety.

📈 Stock Price Action — Five-Year Event Map

CoStar has round-tripped a full valuation cycle. It peaked near $100 in October 2021, spent most of 2022–2024 between roughly the mid-$50s and high-$90s, reached $96.83 in August 2025, and then collapsed to a five-year low of $27.14 on July 23, 2026. The September 1 close of $31.98 is about 68% below the five-year peak and only 6.6% above the roughly $30 reference price in the July memo. The violent rerating happened before the Q2 profitability milestone; the subsequent recovery has been small.

Prices are split-adjusted. Event attribution is interpretation based on earnings and corporate-event dates; it is not proof that any single item caused a given move.

1. Executive Summary

CoStar is best understood as a portfolio of information and marketplace businesses with radically different competitive positions. The flagship CoStar product is a deeply embedded commercial-real-estate database and workflow. LoopNet is a leading commercial-property marketplace. Apartments.com is a powerful multifamily advertising network. Homes.com is an ambitious but still subsidized U.S. for-sale challenger. Domain and OnTheMarket are international residential challengers; Matterport adds spatial data and 3D capture; Zonda adds new-home intelligence, builder workflows and marketplaces. Segment reporting compresses these distinct economics into “Commercial Real Estate” and “Residential Real Estate,” which obscures both the quality of the commercial core and the continuing uncertainty inside residential.

First-half mix illustrates that reporting problem. CoStar product contributed $668 million, LoopNet $172 million and other commercial products $113 million, for $953 million of Commercial revenue. Residential contributed $869 million, but approximately $131 million came from acquired Domain. The 32% reported Residential growth rate therefore combines organic Apartments.com and Homes.com growth with a major acquisition; it should not be read as a clean measure of U.S. portal momentum. Nor is “Commercial” purely subscription data: Ten-X transaction revenue and Matterport sit alongside the more predictable CoStar and LoopNet engines. A sensible forecast begins at the product level, then reconciles to segments.

This mix also changes the interpretation of margin recovery. Commercial’s mature earnings provide the base, Apartments.com likely contributes substantial profit inside Residential, and Domain added revenue and earnings after acquisition. Homes.com is the marginal swing factor, but the company does not disclose it as a standalone profit center. Investors can observe the consolidated cost reset and infer direction; they cannot directly measure whether Homes itself crossed contribution breakeven. That opacity is why a single positive segment quarter deserves less weight than recurring cohort and cash evidence.

The commercial evidence is strong. Q2 2026 Commercial revenue increased 8% to $481 million and adjusted EBITDA reached $172 million, a 36% margin. CoStar product revenue rose 9% to $337 million, subscribers increased 19% to 327,000 and quarterly renewal was 93%. LoopNet revenue advanced 14% to $87 million. These figures arrived while commercial-property financing remained restrictive and development demand was soft, which suggests the subscription engine is more resilient than transaction volumes. The July 2026 Federal Reserve lending survey found some easing in standards but still-tight historical conditions and little overall improvement in CRE loan demand. Federal Reserve July 2026 survey

Residential is not one thing. Apartments.com remains a strong network: Q2 revenue grew 9% to $318 million, paid properties exceeded 92,000 and monthly renewal was 99%. Yet average revenue per property fell about 3.6% as new customers skewed smaller, management discussed aggressive competitor discounting, and Zillow reported 31% Rentals growth and 42% multifamily growth. The combination implies excellent customer continuity but less pristine pricing power than the headline renewal rate suggests. Zillow Q2 results

Homes.com delivered the quarter’s most important positive and its most important ambiguity. Revenue increased 66% to $28.5 million, annualized run rate reached $116 million, subscribers more than doubled year over year to above 36,000, and June monthly cancellations fell to 2.4% from 6.5%. Those gains are real. But sequential net additions were only about 825, and 2.4% monthly cancellation still equates to roughly 25% annual attrition if sustained. Simple multiplication of $305 June average subscriber price by more than 36,000 subscribers and twelve months produces about $132 million, not the disclosed $116 million run rate. Differences in timing, promotional periods or KPI definitions may explain the gap, but investors cannot calculate lifetime value or payback without gross additions, cohort retention and acquisition cost. Management reduced the inside-sales force from roughly 660 at year-end 2025 to about 400, held bookings merely level with Q1, and shifted spending toward lower-funnel digital and search marketing. Efficiency improved; proof of scale did not.

Financially, Q2 marked a real reversal. Consolidated revenue was $925 million, gross profit was $728 million, operating income was $76 million, net income was $55 million, EBITDA was $157 million and adjusted EBITDA was $184 million. For the first half, operating cash flow was $267 million and capital expenditures were $112 million, producing approximately $155 million of conventional free cash flow before acquisitions. A $109 million litigation settlement depressed reported cash flow; normalizing only that item would put first-half free cash flow near $264 million. Stock compensation was $80 million, so owner cash flow after stock compensation was approximately $75 million as reported or $184 million after the litigation normalization. This distinction matters: adjusted EBITDA is useful for tracking the expense reset, but it is not distributable cash.

Capital allocation is the pivotal governance question. CoStar repurchased 13.8 million shares for $589 million in the first half—an average near $42.70, above the current quote—while cash fell and acquisitions continued. The August Zonda acquisition can strengthen CoStar’s data in a valuable adjacent category, but it also extends residential and homebuilding-cycle exposure and consumes liquidity before the returns on Domain and Matterport are separately observable. Founder-led ambition built the commercial franchise; the same ambition can overextend it.

The central analytical conclusion is therefore two-sided. CoStar’s core moat is broad enough to protect recurring commercial cash flows but narrower than the prior memo implied: Moody’s, MSCI and Altus own strong positions in other CRE datasets, indexes and valuation workflows. Apartments.com is a high-quality marketplace facing faster-growing rivals. Homes.com is a moat under construction. Residential’s adjusted breakeven validates management’s ability to cut costs, not yet the proposition that acquired customers earn attractive returns. At the current capitalization, the embedded expectation is no longer catastrophic failure; it is a steady recovery toward materially higher margins. The next two quarters must show that revenue can keep compounding after the sales-and-marketing reset.

2. Business Overview

The commercial information engine

CoStar product is the economic crown jewel. It packages property inventory, lease activity, sale comparables, tenant information, lender data, market analytics and workflow tools into subscriptions used by brokers, owners, lenders, appraisers, investors and public agencies. CoStar says its products cover more than 8.5 million properties, 7 million lease activities, 5 million sale comparables and 8.2 million commercial tenants. The important point is not any single database count; it is the collection system behind those counts. Field researchers, telephone verification, public records, contributed datasets and normalization over decades create a longitudinal record that becomes more useful when an entire team uses the same definitions. CoStar products 2025 Form 10-K

The subscription model prices by product, users, organization and geography. Most contracts renew automatically, and the product becomes part of recurring tasks: sourcing comparables, underwriting a loan, preparing a broker opinion, tracking tenants or monitoring asset performance. That creates workflow and social switching costs rather than hard technical lock-in. An occasional user can switch to public records or another vendor; a national brokerage must retrain teams, reconcile definitions and accept gaps in history. Captivity is therefore strongest among multi-seat institutions and weakest among infrequent users.

New modules can deepen the advantage. Management says Rent Benchmark draws on AI abstraction of four million actual lease documents. Debt Solutions receives anonymized information from 300 lenders covering 100,000 active loans with $1.2 trillion of balances and generated more than $4 million of monthly net-new bookings in Q2. Those claims come from management and should be treated as such, but the mechanism is attractive: contributed leases and loan portfolios are harder to reproduce than scraped public listings, while each participant can improve the aggregate benchmark.

Commercial marketplaces and transaction products

LoopNet monetizes commercial listings and audience. Its economics resemble a specialized marketplace: inventory attracts tenants and investors; traffic attracts brokers and owners; paid prominence improves monetization. Q2 revenue was $87 million, up 14%, and paid listings rose 9% to about 220,000. LoopNet shares CoStar data, imagery and sales relationships, but its consumer acquisition and advertising economics differ from the core subscription database.

Other commercial products include STR hospitality benchmarking, Ten-X auctions, BizBuySell and Land.com. They are not equally strong. STR benefits from contributed hotel-performance data. Ten-X is transaction-sensitive and weakened enough that restructuring it accounted for roughly one-quarter of the 2026 revenue-guidance reduction. “Other commercial” revenue fell 5% in Q2 even as Matterport subscriptions grew 16%. The flagship product’s moat does not automatically apply to every brand.

Apartments.com

Apartments.com is the second established franchise. Property managers pay to promote communities across a rental network, and renters search without paying. The 99% monthly renewal figure, more than 92,000 paid properties and 228 million management-reported Q2 renter visits indicate real two-sided liquidity. Its advantages are brand, inventory breadth, search visibility, a national salesforce and integration with CoStar’s multifamily data.

The caveat is competitive supply. Zillow Rentals reported 79,000 multifamily properties and 2.8 million monthly rental listings while growing substantially faster in Q2. The Federal Trade Commission’s August order unwinding parts of Zillow’s Redfin rentals agreement removes an inventory shortcut for Zillow, but it also requires Redfin to rebuild an independent rental marketplace within six months and invest millions in that effort. Near term, contract reopening may help Apartments.com win accounts; structurally, another funded marketplace increases supply. FTC Zillow/Redfin case

Homes.com and international residential portals

Homes.com sells memberships to agents under a “your listing, your lead” proposition. Unlike lead-auction models that route a listing’s consumer inquiry to an advertising buyer agent, CoStar promises the listing agent the lead and promotes that agent’s inventory. Members promoted about 305,000 active listings in Q2, 9.3% of 3.2 million U.S. listings. This concentration means current customers appear more productive than the median agent, which is useful for monetization but makes the total licensed-agent count a misleading addressable-market denominator.

The realistic customer pool is agents and teams with enough listings and commission income to justify a recurring marketing expense. The National Association of Realtors reported 1.44 million members in 2026, but the typical member completed only nine transaction sides and earned $59,200 of gross income in 2025; agents with two years or less experience earned only $8,000. At $305 per month, Homes costs $3,660 annually—6.2% of median gross income and nearly half the novice figure before other business expenses. That does not prevent a large business, but it makes productive listing agents, teams, brokers and depth advertising the more relevant opportunity set. NAR 2026 Member Profile

Domain in Australia and OnTheMarket in the United Kingdom expand the residential portfolio but begin from challenger positions. Domain’s pro-forma Q2 revenue grew 9% and management reported 41 million monthly visits. REA reported 150 million monthly visits, 12% residential-revenue growth and a 62% first-half EBITDA margin. OnTheMarket’s inventory and leads are growing, yet Rightmove reported 90% of portal time and a 69% underlying operating margin. These are credible footholds, not replicas of CoStar’s U.S. commercial dominance. REA Q3 update Rightmove H1 results

Matterport and Zonda

Matterport provides spatial capture, 3D tours and digital-twin data that can differentiate listings and supply structured visual information to CoStar products. The commercial logic is intuitive: more complete digital records can improve engagement and reduce creation costs across Apartments.com, Homes.com, LoopNet and Domain. The economic return is not yet visible because Matterport is not separately reported and integration costs are mixed with the broader platform.

Zonda adds proprietary lot-level and new-construction data, builder and lender workflows, and consumer marketplaces including NewHomeSource and Livabl. More than 3,000 customers and a 23% adjusted-EBITDA margin establish that this is an operating business rather than a pre-revenue option. Its data can feed Homes.com and CoStar’s new-homes coverage; its customer relationships can support cross-selling. It also increases exposure to construction volumes and creates a fourth major integration alongside Domain, Matterport and international CoStar launches.

3. Industry Structure and Capital Cycle

Commercial real-estate information

Commercial property information has favorable industry structure. Coverage and accuracy require high fixed costs: researchers, normalization, imagery, public-record ingestion, software and years of history. Once those costs are incurred, distributing another subscription is inexpensive. A scaled provider can spend more per market while charging less per datapoint than a new entrant. Customers also benefit when colleagues, counterparties and lenders refer to the same comparable-sales and tenant records. Economies of scale and customer captivity reinforce each other.

This does not make CoStar a monopoly over all CRE information. Moody’s says its platform covers more than eight million commercial properties with four decades of history; MSCI offers more than $50 trillion of transaction data and more than 90 real-estate indexes; Altus ARGUS is embedded in valuation and cash-flow modeling. CoStar is strongest in U.S. broker, owner, property, lease and tenant workflows. MSCI is strong in institutional transactions and benchmarks, Moody’s in credit and risk, and Altus in valuation software. The prior claim of roughly 90% market share across “CRE information” was too sweeping. The defensible conclusion is a wide, segment-specific moat.

Capital-cycle conditions remain favorable in that segment. High returns have not produced a comparably complete new nationwide property-and-comps platform because the entrant would fund years of collection before achieving customer trust. Competition tends to form around adjacent workflows or narrow datasets, which may cap pricing but does not instantly erase the installed information standard. Artificial intelligence lowers interface and extraction costs, yet it cannot conjure verified private lease terms or decades of normalized transactions. AI is more likely to raise the value of proprietary inputs while commoditizing generic summaries.

Rental and for-sale portals

Residential portals have worse supply dynamics. Their high gross margins and visible consumer traffic invite continuous spending by Zillow, CoStar, Redfin, brokerages, search engines and local incumbents. Inventory is widely syndicated, product features are imitable, and consumer attention must be reacquired. Scale can produce network effects, but the market can support multiple interfaces because the same listing appears on several sites. Marketing and search distribution absorb much of the theoretical incremental margin.

Apartments.com has crossed the threshold where consumer habit and inventory reinforce each other. Its renewal and paid-property growth support that view. Still, falling average revenue per property and Zillow’s faster rental growth show that captivity is incomplete. Homes.com has not crossed that threshold. It has inventory and a differentiated agent promise, but only about 2.5% penetration of NAR members and a subscriber base that requires continual replacement at current churn.

Google’s Home Listings Ads illustrate the distribution risk accurately. The current product surfaces for-sale listings in U.S. mobile search, is powered by ComeHome/HouseCanary, promotes buyer agents and charges per lead. It is not a complete free portal and does not cover rentals, commercial property or land. The prior “Google is entering as a full home portal” description overstated the scope. Yet Google controls valuable top-of-funnel queries, and Homes.com’s shift toward search-engine marketing can transfer economics to the distributor. Zillow says 80% of its traffic is direct to apps and sites and has announced exclusive real-estate integration in Google’s Gemini ecosystem, giving it a stronger distribution hedge. Google Home Listings Ads Zillow Q2 shareholder letter

The capital-cycle verdict is therefore bifurcated. In core CRE information, fixed-cost research and workflow standards restrict effective supply. In residential portals, attractive incumbent margins cause more capital, more advertising inventory and more interfaces to enter. CoStar itself is part of that supply response. Residential can become valuable, but its equilibrium return should be assumed lower than the mature commercial database until customer economics prove otherwise.

4. Competitive Position and Moat

CoStar product: wide moat, carefully defined

The moat mechanism has four layers. First is replacement cost: assembling and continuously checking the dataset would require a large organization and years of loss-making investment. Second is density: broad coverage lowers the probability that a customer must use a second source. Third is longitudinal consistency: a long time series matters for comparisons, underwriting and appraisal. Fourth is workflow captivity: teams train on the product, exchange its records and embed it in recurring processes.

The operating evidence is compatible with a wide moat. Quarterly product renewal of 93% remained high in a weak property market. Subscriber growth of 19% outpaced revenue growth of 9%, indicating broader adoption but also a mix or monetization question. Commercial adjusted margin held 36% despite international investment and Ten-X weakness. Company-wide trailing-twelve-month renewal for contracts of at least a year was 89%, unchanged year over year. These are strong—not invulnerable—retention figures.

There are three limits. Customer captivity is contractual and procedural, not technological. Data quality must be maintained continuously; stale records would erode trust quickly. And rivals can dominate adjacent categories without rebuilding the entire platform. A lender may use CoStar property data, Moody’s credit analytics, MSCI indexes and ARGUS models in the same workflow. CoStar’s opportunity is to capture more of that stack through Debt Solutions and new modules; the risk is that bundling invites stronger multi-product competition.

Apartments.com: network effects with contested pricing

Apartments.com’s moat is a two-sided audience rather than a proprietary-data standard. Renters prefer broad, current inventory; property managers pay where renter leads convert. Brand advertising and search rank reinforce the loop. Ninety-nine percent monthly renewal signals low near-term churn, while 12% paid-property growth suggests the network is still expanding.

The weak point is monetization. Average revenue per property declined as smaller communities joined, and management chose to defend pricing while a competitor discounted. That may be rational long term, but it means volume growth does not translate one-for-one into revenue. Zillow’s 31% rental growth and 42% multifamily growth demonstrate that Apartments.com does not own the category. Traffic comparisons from the two companies use incompatible definitions and should not be interpreted as proof of share transfer.

Homes.com: differentiated proposition, unproven captivity

“Your listing, your lead” solves a genuine agent grievance and aligns Homes with listing-side professionals. The product can also draw on CoStar’s content, Matterport tours and large advertising apparatus. Falling cancellations after the price increase are encouraging. The planned Platinum or “depth” products could raise revenue per customer by selling greater prominence, a model CoStar says is common at Apartments.com and Domain.

But depth creates a strategic tension. If paying more changes listing prominence or lead allocation, the product begins to resemble the pay-to-rank systems from which its egalitarian message differentiates it. More important, the company has not disclosed CAC, contribution margin, gross adds, annual cohorts, depth attach or Homes standalone profit. At 2.4% monthly cancellation, about 10,400 gross annual additions would be required merely to hold a 36,000-member base constant. That hurdle may be manageable, but investors cannot verify it from the current KPIs.

International: local network effects travel poorly

CoStar’s commercial research process is portable: hire researchers, assemble normalized data, migrate existing customers and use LoopNet as a demand surface. France and Australia provide credible opportunities, although migrations and data freshness take years. Residential consumer habit is less portable. REA and Rightmove possess local network effects, direct traffic and extraordinary margins; buying the number-two or number-three site does not import the leader’s audience. The evidence supports treating the international portfolio as challenger-grade until revenue and margin convergence appear.

5. Growth History and Forward Opportunities

Revenue increased from $1.66 billion in 2020 to $3.25 billion in 2025, a compound rate near 14%, through organic price-and-seat growth and acquisitions. The mix changed considerably: Domain and Matterport enlarged reported growth, while Homes.com spending suppressed profit. First-half 2026 revenue was $1.822 billion, up 20%, including acquired revenue; Q2 organic trends were more modest.

Highest-quality growth: modules and international commercial data

Debt Solutions, Rent Benchmark and lender workflow expand wallet share among captive users and add contributory datasets. The incremental economics should be attractive because sales relationships, research and infrastructure are shared. International CoStar launches use an established operating model but require patient investment. In France, management reported data on 290,000 properties, 385,000 tenants, 90,000 availabilities and 75,000 comparables, with 1,100 Business Immo subscribers expected to migrate over two years. Australia had 124 researchers and photographers ahead of a second-half 2026 launch, while LoopNet Australia was deferred to late 2027. These are multi-year builds, not immediate margin contributors.

Apartments.com: inventory growth versus yield

Paid-property growth can support high-single-digit revenue even when per-property revenue falls. Over time, however, value requires either stable yield, upsell into higher tiers or lower acquisition and service cost. Competitor discounting tests the platform’s ability to price on lead quality rather than inventory count. The best evidence will be same-property revenue, cohort retention and advertiser return—not visits alone.

Homes.com: price, penetration and depth

Homes has three levers. Standard membership pricing increased in May. Productive-agent penetration can rise from a low base. Platinum/depth advertising can expand yield among members with valuable listings. These levers could grow revenue even with a smaller salesforce if churn stays down and sales productivity improves. The contrary case is equally clear: pricing raises churn after anniversary dates, depth cannibalizes the original proposition, and lower sales capacity makes gross additions insufficient. Q2’s flat sequential bookings leave both paths open.

Acquired growth

Domain, Matterport and Zonda add revenue faster than the legacy business but must be judged on return, not scale. Domain provides an Australian portal and transaction exposure; Matterport provides proprietary spatial data; Zonda provides builder intelligence. Cross-selling is plausible, but the company does not disclose acquired-asset ROIC or standalone margins after corporate allocations. Product logic and demonstrated economic value are separate questions.

Near-term growth bridge

Management’s 2026 revenue guidance of $3.715–$3.755 billion implies roughly 15% growth at the midpoint, while second-half organic growth is expected around 8.5%–9%. Commercial guidance is $1.94–$1.96 billion with a 35% adjusted margin; Residential guidance is $1.775–$1.795 billion with $110–$130 million of adjusted EBITDA. About one-quarter of the revenue-guide reduction relates to Ten-X; the rest is mainly Homes salesforce optimization and Apartments pricing discipline. Those choices improve profit quality only if they do not impair future bookings.

The leading indicator is less reassuring than reported revenue. Q2 net-new bookings were $69 million, down 26% year over year and only 3% sequentially, although commercial CoStar bookings rose 24%. Subscription revenue will lag that deceleration. The next phase of growth must come from productive core bookings, Homes depth adoption and acquired cross-sell rather than the easy comparison against 2025 residential spending.

6. Financial Quality and Earnings Reconstruction

Income statement

The gross-margin structure remains excellent. Q2 gross profit of $728 million on $925 million of revenue equaled 78.7%. Operating income was $76 million, an 8.2% margin, versus a loss a year earlier. The gap between gross and operating margin reflects sales, marketing, product development, research and acquired-company integration—not deterioration in the underlying information product.

The distinction between EBITDA and adjusted EBITDA is not semantic. Q2’s $27 million difference included stock compensation and other adjustments. Residential crossed zero only after those exclusions. For the first half, Residential adjusted EBITDA was approximately negative $17 million and actual EBITDA about negative $51 million. Homes.com profitability is not separately disclosed, so positive Apartments.com and Domain economics may subsidize it within the segment.

Cash conversion

First-half operating cash flow was $267 million. Capital expenditures comprised $83 million for the Richmond campus and $29 million of other capex, leaving approximately $155 million of conventional free cash flow. The Brown litigation settlement produced a $109 million cash outflow; adding only that unusual item produces approximately $264 million. Stock compensation was $80 million. An owner-oriented calculation therefore yields roughly $75 million of reported free cash flow after SBC, or $184 million after the settlement normalization.

This is an improvement from 2024–2025 but remains well below adjusted EBITDA. Some campus spending is genuinely temporary and should fall. Stock compensation is recurring economic compensation even when share repurchases offset dilution. Acquisition payments are excluded from free cash flow by convention but are central to long-run owner returns for a serial acquirer.

Balance sheet and pro-forma liquidity

At June 30 CoStar had $1.266 billion of cash and approximately $994 million of long-term debt, before the Zonda close. Static subtraction of the $800 million purchase price leaves about $466 million of cash and roughly $528 million of net debt, excluding post-quarter cash generation, transaction costs, acquired cash or working-capital adjustments. This is an estimate, not a reported post-close balance sheet. Leverage remains manageable relative to commercial earnings, but the earlier net-cash cushion is largely gone.

Goodwill and acquired intangibles were approximately $6.65 billion before Zonda, making book value a weak measure of downside protection. Impairment testing does not establish that acquisitions earned their cost of capital; it only tests whether carrying values remain recoverable under accounting assumptions. The balance sheet is solvent, but its asset quality depends on management’s deal underwriting.

Returns on capital

Consolidated historical ROIC is distorted downward by residential spending, campus capex, idle acquisition goodwill and cash. A segment lens is more informative. Commercial’s 35%–36% adjusted margin, recurring revenue and limited incremental working capital imply attractive returns on the mature data base. Residential’s return remains unknowable because Homes acquisition costs and profit are not separated from Apartments.com, Domain and shared expenses. The correct conclusion is not that CoStar’s entire franchise has lost its moat; it is that management is reinvesting high-return core cash in projects whose returns have not yet been demonstrated.

Guidance quality

Holding EBITDA guidance while cutting revenue guidance shows credible cost control but lowers the quality of the bridge. Management plans roughly $100 million of expense reductions, including sales and marketing. A margin gained by eliminating ineffective spending is valuable; a margin gained by starving customer acquisition is temporary. The unresolved question is whether the Q4 Residential profit ramp can be achieved without another booking slowdown. Q2 management did not provide enough product-level detail to reconcile that bridge.

7. Capital Allocation, Management and Governance

CoStar’s capital-allocation record has two identities. The first is admirable: founder and chief executive Andrew Florance spent decades reinvesting in a proprietary research operation, bought useful datasets and marketplaces, and created a commercial-information franchise with high recurring revenue and attractive incremental economics. The second is more difficult: a willingness to pursue several large adjacencies at once, accept years of consolidated margin compression, use adjusted measures generously and continue acquisitions before earlier deals have produced separately visible returns.

Reinvestment

The best reinvestment is extension of CoStar’s existing data advantage. More researchers, better normalization, lender-contributed data, international coverage and modules sold to the existing customer base can raise the value of the core network. These projects exploit an established sales channel and information asset. Their risk is execution, not the need to create consumer habit from zero.

Homes.com is different. The project uses commercial-core cash to acquire consumer traffic and agent subscribers in an arena with an entrenched leader. Management reduced its annualized spending trajectory in Q2 and reported better sales productivity, but cumulative spending remains large relative to Homes revenue. The question is not whether management can make the segment cross an adjusted accounting threshold; it is whether lifetime gross profit from retained customers exceeds acquisition, service and continuing traffic costs. Current disclosure does not permit that calculation.

Campus capital expenditures are another allocation. Richmond spending was $83 million in the first half, with $29 million of other capex. The campus can support research and operations, but buildings do not create the same scalable return as data. As the project winds down, reported free cash flow should improve mechanically. That cash-flow recovery should not be mistaken entirely for operating progress.

Acquisitions

CoStar has used acquisitions to enter adjacent categories faster than organic development would allow. Matterport added spatial capture; Domain added a major Australian portal; Zonda adds new-home data and builder relationships. Each has strategic logic. The portfolio problem is simultaneous integration: product teams, sales incentives, data architectures, international launches and brand spending all compete for management attention.

The $800 million Zonda price implies about 4.7 times 2025 revenue and 20.5 times adjusted EBITDA. At that starting yield, value creation requires growth, synergies or both. Expected adjusted-EPS accretion in the first full year is not sufficient evidence because acquisition accounting, financing and excluded costs can make adjusted EPS rise even when economic returns are below the cost of capital. The useful tests are retained organic growth, cash conversion, customer cross-sell and return on the full purchase price.

The balance-sheet sequence is important. CoStar entered Q2 with substantial liquidity, used $589 million for buybacks in the first half, and then committed $800 million to Zonda. The static post-close net-debt estimate is modest, but the capacity to fund another large deal from cash has narrowed. This may impose welcome discipline; it may also shift future expansion toward debt or equity if management does not slow acquisition activity.

Repurchases and insider signals

The first-half repurchase retired 13.8 million shares and reduced Q2 diluted shares to 404.4 million from 424.3 million a year earlier, a decline of about 4.7%. Buying below long-term historical valuations can create value, but the average first-half purchase price near $42.70 exceeded the September quote. More fundamentally, repurchases exceeded free cash flow by a wide margin and were followed by the cash-funded acquisition. A durable program should be funded by recurring owner cash flow after stock compensation and required capital spending.

Florance bought 83,300 shares in the open market on August 4 at an average $29.89, approximately $2.49 million, and the transaction was not marked as a prearranged Rule 10b5-1 trade. That is a useful alignment signal because it uses personal capital after the Q2 release. It does not resolve the business economics; insiders can be early or wrong. August 2026 Form 4

Governance and measurement

The board formed a Capital Allocation Committee following its 2025 settlement with Third Point, an institutional response to shareholder concern about residential spending and executive compensation. The committee’s practical effectiveness should be judged by disclosure and capital outcomes, not its existence.

The 2026 proxy raises a measurement concern. It reports a 200% payout on $299.784 million of compensation-plan “EBITDA” against a $278.112 million target, while the 2025 filing reports approximately $170 million of EBITDA and the earnings release reports $442 million of adjusted EBITDA. These may be legitimately different definitions, periods or incentive adjustments, but the proxy does not provide a bridge that reconciles them. When management compensation relies on a bespoke measure between GAAP EBITDA and public adjusted EBITDA, shareholders cannot readily test pay-for-performance. 2026 proxy statement

The governance verdict is balanced. Founder ownership and continued open-market buying align Florance with shareholders, while his long record supports the ability to build difficult data assets. The same concentration of strategic authority and acquisition appetite creates key-person and empire-building risk. The Capital Allocation Committee, clearer segment reporting and restraint after Zonda would improve confidence; another large deal before product-level returns emerge would weaken it.

8. What Changed Since July 3, 2026

The update contains four positive changes, six adverse changes and three important corrections.

Positive evidence

1. Residential crossed adjusted breakeven. The $12 million Q2 result passed the prior memo’s immediate test on management’s preferred metric, versus losses of $76 million a year earlier and $29 million in Q1.
2. Homes retention and price improved. June cancellation fell to 2.4%, annualized run rate reached $116 million and the May price increase did not trigger an immediate visible reversal.
3. The commercial core held. CoStar product subscribers rose 19%, renewal was 93%, revenue grew 9% and Commercial adjusted margin was 36%.
4. Insider alignment increased. Florance invested another $2.49 million personally after the Q2 release.

Adverse evidence

1. Growth expectations fell. Full-year revenue guidance was reduced by $65 million at each end, and net-new bookings declined 26% year over year.
2. The Residential milestone was narrow. Plain EBITDA was still negative, H1 Residential remained loss-making, and Homes standalone profit remains undisclosed.
3. Customer acquisition slowed. Homes added only about 825 subscribers sequentially, bookings were merely similar to Q1, and the salesforce was cut roughly 39% from year-end.
4. Apartments pricing softened. Paid properties grew faster than revenue, revenue per property fell and Zillow’s rental business grew faster.
5. Capital allocation expanded again. Zonda consumed $800 million of cash and added integration and housing-cycle exposure before Matterport and Domain returns were isolated.
6. The finance organization changed during the integration wave. Christian Lown resigned for an outside-industry opportunity, and international managing director Robin Rossmann became chief financial officer effective July 31. The filing says the departure did not involve a disagreement, but a CFO transition while guidance, cost reductions and three large integrations are moving at once adds execution risk. CFO transition 8-K

Corrections to the prior memo

The old description of a roughly 90%-share CRE-information monopoly is unsupported at that breadth. CoStar has a wide moat in specific property, lease, tenant and broker workflows, while Moody’s, MSCI and Altus hold important adjacent positions. The old description of Google’s move as a full home-listings portal also overstated the current product. Google’s offering is presently a U.S. mobile, for-sale lead-ad product, though its control of search still creates distribution risk. Finally, CoStar’s 10-for-1 stock split occurred on June 28, 2021, not in 2024.

Market and factor context

The shares rose 17.8% from the July 23 five-year low of $27.14 to $31.98 but remain 64.3% below the trailing 52-week high and 67.9% below the five-year peak. They sit about 1.7% above the 50-day exponential moving average and 26.5% below the 200-day average; the 50-day remains 27.7% below the 200-day. That is short-term stabilization inside an unrepaired primary downtrend, not a technical recovery.

Raw adjusted-close returns were positive 6.5% over one month, negative 5.0% over three months, negative 31.6% over six months and negative 64.3% over twelve months. A broad factor model through July 31 explained only about one-third of return variation and showed positive market, real-estate and software/data exposure, negative momentum exposure and only a small value loading. The drawdown is therefore not reducible to one macro factor. Company-specific residential spending and expectation resets matter alongside broader duration and property sensitivity.

9. Risk Analysis

Search and AI

Search risk is not simply a decline in website visits. A search or AI platform can place its own answer, listing carousel or paid lead product above organic results, increasing acquisition cost and weakening the portal’s bargaining position even if total consumer demand is unchanged. Homes is particularly exposed because management is reallocating spend toward search-engine marketing while building direct habit. Apartments.com and Zillow have stronger existing brands, and Zillow says four-fifths of its traffic is direct.

AI also creates opportunities. Proprietary lease, loan, property and spatial data can support better answers than public-web scraping. CoStar can integrate natural-language search into captive professional workflows, where reliability and provenance matter. The risk is asymmetrical across the portfolio: AI may enhance the commercial data asset while commoditizing the consumer portal interface.

Cyclicality

CRE subscription revenue is less cyclical than property transactions but not immune. Broker layoffs, customer consolidation and weak development can reduce seats. Ten-X and transaction products react faster. July bank data suggested standards were easing from restrictive levels while demand remained largely unchanged and construction lending weakened. A recovery could improve bookings and auction volumes; a prolonged refinancing stress could pressure smaller customers.

Residential advertising depends on agent and property-manager economics. Low transaction volumes can raise the need for leads but reduce budgets and customer survival. Homes.com’s target customer is likely more productive than the median agent, offering some protection. Depth advertising could become more valuable when listings are scarce, or harder to justify when commissions and closings fall.

Accounting and disclosure

The central accounting risk is not fraud; it is analytical opacity. Segment adjusted EBITDA mixes mature and experimental businesses. SBC is excluded from the preferred profit measure. Acquisition costs and intangible amortization are treated as non-core even though acquisitions recur. Management discloses attractive operating KPIs but not enough cohort data to calculate Homes economics. The most reliable anchors are GAAP operating income, cash flow after capex and SBC, share count, and product-level revenue and retention where available.

10. Valuation and Embedded Expectations

Current capitalization

At the September 1 close of $31.98 and 405.2 million latest reported shares, market capitalization is approximately $12.96 billion. Using June 30 long-term debt of about $994 million, $1.266 billion of cash and roughly $13 million of noncontrolling interests produces pre-Zonda enterprise value near $12.70 billion. Subtracting the $800 million cash purchase price on a static basis produces post-Zonda cash near $466 million and enterprise value around $13.50 billion. Actual post-close cash will differ because of third-quarter cash generation, deal costs, acquired cash and working capital.

Against trailing pre-Zonda revenue of about $3.56 billion, static post-close enterprise value is roughly 3.8 times sales. Adding Zonda’s $170 million of 2025 revenue gives about $3.73 billion of pro-forma revenue and a 3.6-times multiple. Against the $800 million midpoint of CoStar’s 2026 adjusted-EBITDA guidance plus Zonda’s approximately $39 million of 2025 adjusted EBITDA, the multiple is around 16 times. On a mixed-basis trailing check—CoStar’s $394 million of actual EBITDA plus Zonda’s disclosed adjusted EBITDA—the multiple is about 31 times. This is not a fully pro-forma forecast: CoStar guidance may include none or only a partial period of Zonda, and synergies, transaction costs and acquired amortization are excluded. It is a capitalization sanity check.

Trailing operating cash flow was about $497 million and capex $270 million, leaving $227 million of reported free cash flow. Adding back only the $109 million Brown settlement produces $336 million; after $192 million of trailing SBC, the corresponding owner-cash figures are only $35 million and $144 million. Static post-Zonda enterprise value is therefore about 59 times reported FCF or 40 times litigation-normalized FCF before deducting SBC. The low sales multiple coexists with expensive current cash production.

Historical and peer context

CoStar’s sales multiple compressed from roughly 16 times at year-end 2021 and 14 times at year-end 2022 to about 8.6 times at year-end 2025 and roughly 3.7 times at the current price. The stock’s price-to-sales and price-to-book readings are near the bottom of their own history, while the earnings multiple is meaningless because GAAP earnings are depressed. Historical compression is informative about sentiment, not intrinsic value. The earlier multiples capitalized 25%–30% margins, low rates and confidence in reinvestment; today’s company has a different mix, more goodwill and unresolved residential economics.

Peer comparisons require care. Wide-moat information providers such as Moody’s, S&P Global and Verisk earn far higher current margins and deserve higher sales multiples. FactSet and Gartner broaden the workflow-information set, while Zillow is only a residential-portal cross-check because its monetization and stock compensation differ. A secondary market screen near September 1 put the five core peers at a median 8.8 times sales and 17.4 times EBITDA; Zillow was about 2.6 times sales and 42 times EBITDA. CoStar’s roughly 3.6 times pro-forma sales is well below the core median, but its 31-times trailing actual-earnings cross-check is above every core peer. The 16-times forward adjusted multiple looks peer-like only if guidance is delivered. S&P Global statistics Moody’s ratios Verisk statistics FactSet statistics Gartner statistics

Earnings-power value

A Greenwald-style earnings-power approach asks what the current business can earn without assuming heroic growth. Commercial guidance of approximately $1.95 billion of revenue at a 35% adjusted margin implies around $683 million of adjusted EBITDA. Residential guidance of about $1.785 billion and $120 million of adjusted EBITDA adds a 7% margin, for $803 million consolidated before the Zonda contribution. Converting that figure to owner earnings requires taxes, recurring capex, cash interest, working capital and SBC; it also requires deciding how much international and product spending maintains rather than grows the franchise.

If normalized after-tax owner earnings were only $400–$500 million, the current enterprise value would capitalize them at roughly 27–34 times. If cost normalization and campus completion raise owner earnings toward $600–$700 million without sacrificing growth, the multiple falls toward 19–23 times. The market therefore does not need Homes.com to become Zillow, but it does need the reported margin recovery to convert into cash and persist. Current price cannot be justified solely by today’s GAAP earnings.

Reproduction value offers a different check. Rebuilding CoStar’s decades of normalized property, lease, tenant and transaction history would require many years of research spending, customer losses during ramp and uncertain adoption; accounting book understates that internally built asset. Conversely, the $6.65 billion of goodwill and intangibles before Zonda records prices paid for acquired assets, not proof of economic reproduction value. The commercial database plausibly merits a premium to tangible capital, while acquired residential portals should be marked by their demonstrated cash generation rather than purchase cost.

Reverse embedded expectations

One way to frame the current enterprise value is as a steady-state cash-flow requirement. At an 8.5% cost of capital and 3% perpetual growth, $13.5 billion of enterprise value corresponds to approximately $743 million of sustainable after-tax free cash flow before excess non-operating adjustments. At a 9.5% cost and 2.5% growth, the requirement is about $945 million. These are simplified perpetuity calculations, not forecasts. A ten-year reverse cash-flow model using a 9% discount rate and 3% terminal growth requires about 13.6% annual growth from litigation-normalized $336 million FCF, or about 24.9% from the $144 million figure after SBC. Compared with reported first-half owner cash flow, both require substantial further conversion; compared with commercial adjusted EBITDA plus a normalized Residential contribution, the lower figure is attainable but not yet demonstrated.

The valuation embeds three practical expectations:

1. Commercial revenue continues high-single-digit growth with margins around the mid-30s rather than suffering material renewal or pricing erosion.
2. Residential remains profitable after adjustment and moves toward actual EBITDA and cash profitability without renewed marketing escalation.
3. Capex falls as campuses complete, stock compensation is contained, and acquisitions do not absorb the cash recovery.

Failure of one can be offset by strength in another, but failure of two would make the current capitalization difficult to support from cash flows.

Operating scenarios—not price targets

These cases isolate operating outcomes. Applying a valuation multiple without specifying interest rates, reinvestment needs and acquisition policy would create false precision. The critical asymmetry is that margin normalization can produce rapid earnings growth even with moderate revenue growth, while renewed portal spending can reverse that leverage just as quickly.

11. Variant Perception

What the market may be missing positively

The commercial core is more durable than consolidated margins imply. Its 36% adjusted margin and 93% renewal survived a weak property cycle while subscriber count expanded. If campus capex and ineffective residential marketing fall together, cash flow can rise much faster than revenue. A portfolio view may also underappreciate contributory data: actual leases, lender portfolios, spatial models and builder-lot data can deepen the proprietary information graph across products.

The market may also treat Residential as synonymous with Homes.com losses. Apartments.com is a mature, high-retention network; Domain and Zonda have real revenue; Homes itself has crossed $100 million of annualized revenue. A cost reset that preserves retention would reveal more existing earnings power than the segment’s history suggests.

What optimistic investors may be missing

Adjusted breakeven can be manufactured faster than a marketplace moat. Cutting salespeople and advertising produces immediate EBITDA, while lower gross additions appear later in revenue. Flat sequential Homes bookings, minimal subscriber additions and the full-year revenue cut are consistent with that risk. Monthly churn of 2.4% is a large improvement but still requires a substantial replacement engine.

The usual sum-of-the-parts argument also overstates separability. Commercial cash funds residential investment, shared traffic and data support multiple brands, and management has repeatedly chosen acquisition and expansion over harvesting the core. A hypothetical sale or shutdown value is less relevant when governance indicates continued reinvestment. After Zonda, residential is not simply an option with no balance-sheet cost.

Finally, multiple mean reversion is a weak thesis. CoStar’s former 14–20-times sales valuation reflected a different margin, capital structure and market regime. The relevant question is whether future owner cash flow earns an adequate return on today’s enterprise value, not whether the chart revisits an old multiple.

The differentiated synthesis

The variant is neither “the core is broken” nor “the portal is free.” The core is a wide-moat, high-return information utility whose cash flow has been deliberately redirected into lower-certainty adjacencies. Q2 proved management can reduce the subsidy. It did not prove the adjacency has become self-reinforcing. The most important next datapoint is not traffic; it is the relationship between Homes gross additions, churn, price and acquisition cost after the salesforce reset.

12. Fact Versus Interpretation

13. Open Questions for Management and Diligence

1. What were Homes.com gross additions, net additions and cancellations by monthly cohort after the May price increase?
2. How are “average subscriber price,” reported subscribers and annualized revenue run rate defined, and why does simple multiplication not reconcile?
3. What is fully loaded Homes customer-acquisition cost, including sales compensation, brand media, search marketing and onboarding?
4. What percentage of Homes members renew at the first annual anniversary, and how does renewal vary by agent productivity and geography?
5. How much of Residential’s Q2 adjusted EBITDA came from Apartments.com, Domain, OnTheMarket and Homes individually?
6. What gross-add and depth-product assumptions underpin the steep Residential EBITDA increase implied for Q4?
7. Does Platinum/depth advertising alter lead routing or ranking in a way that changes “your listing, your lead”?
8. What caused Apartments.com revenue per paid property to decline, and what is the same-property revenue trend?
9. How much traffic and paid acquisition for each major portal comes from Google search, other search, direct/app and AI referrals?
10. What organic revenue growth and cash contribution did Matterport and Domain produce after integration costs in 2026?
11. What revenue retention, cash margin and cross-sell hurdles did the board use to approve the Zonda purchase?
12. How will management report Zonda so investors can distinguish acquired growth from organic Homes and commercial growth?
13. Will future repurchases be capped by owner free cash flow after SBC and required capex?
14. How does the proxy’s $299.784 million incentive “EBITDA” reconcile to filed EBITDA and public adjusted EBITDA?
15. What portion of the remaining campus capex is committed, and what is normalized maintenance capex after completion?
16. How sensitive are CoStar seats and renewal to broker headcount, customer consolidation and CRE transaction volumes?
17. What data rights govern contributed lease and lender information, and how does the company mitigate AI-output and privacy liability?
18. Will the Capital Allocation Committee publish acquisition return thresholds or post-investment scorecards?

14. What Must Be True—and Falsifiers

Conditions required for the constructive operating case

• Commercial renewal remains around 90% or better, revenue sustains high-single-digit growth and adjusted margin remains in the mid-30s.
• Homes cancellation stays near or below 2.5% monthly while gross additions reaccelerate enough to grow subscribers after the salesforce reduction.
• Residential produces positive plain EBITDA—not only adjusted EBITDA—for at least two consecutive quarters.
• Apartments.com restores stable revenue per property while maintaining paid-property growth and high renewal.
• Full-year adjusted EBITDA converts into materially higher operating cash flow after recurring capex and stock compensation.
• Domain, Matterport and Zonda show separately observable organic growth or cross-sell without another major acquisition.
• Net debt remains manageable and share repurchases no longer outrun owner cash flow.

Bull-case falsifiers

• Residential returns to adjusted losses after Q2 or remains negative on plain EBITDA through the first half of 2027.
• Homes run-rate growth slows below the pace of price increases, implying stagnant or shrinking real customer volume.
• Monthly cancellation rises materially above 3%, gross additions remain undisclosed, or first-anniversary cohorts deteriorate.
• Commercial renewal falls below the high-80s, subscriber growth fails to monetize, or margin falls despite reduced investment.
• Apartments paid-property growth continues while revenue per property deteriorates enough to hold growth below mid-single digits.
• Management announces another large acquisition before providing return evidence for Domain, Matterport and Zonda.

Bear-case falsifiers

• Homes sustains strong net subscriber additions after the smaller salesforce, depth products gain adoption without higher churn, and standalone economics are disclosed.
• Residential plain EBITDA remains positive for multiple quarters while consolidated bookings reaccelerate.
• Owner cash flow after SBC approaches adjusted earnings as campus capex declines, enabling buybacks without additional leverage.
• CoStar modules and international launches keep Commercial growth in double digits without margin dilution.
• Search/AI referral economics remain benign or proprietary-data integrations make CoStar a preferred upstream provider rather than a disintermediated portal.

Monitoring dashboard

15. Public Source Appendix

Primary company and SEC materials

• CoStar 2025 Form 10-K, filed February 18, 2026
• CoStar Q1 2026 earnings release, filed April 28, 2026
• CoStar Q2 2025 Form 10-Q
• CoStar Q2 2025 earnings release
• CoStar Q2 2026 Form 10-Q, filed July 29, 2026
• CoStar Q2 2026 earnings release, July 28, 2026
• CoStar Q2 2026 earnings event and webcast, July 28, 2026
• CoStar 2026 proxy statement
• Zonda acquisition closing 8-K, August 21, 2026
• Zonda acquisition closing release
• Andrew Florance Form 4, filed August 5, 2026
• Chief financial officer transition 8-K, filed July 13, 2026
• CoStar product overview and database statistics

Competitors, regulators and industry evidence

• Zillow Q2 2026 financial results
• Zillow Q2 2026 shareholder letter
• FTC Zillow/Redfin rentals case and order, August 2026
• Google Home Listings Ads help page
• Google Search and AI Mode update, May 19, 2026
• Federal Reserve July 2026 Senior Loan Officer Opinion Survey
• National Association of Realtors 2026 Member Profile
• REA Group Q3 FY2026 update
• REA Group H1 FY2026 results
• Rightmove H1 2026 results
• Moody’s commercial-real-estate solutions
• MSCI Real Assets market data

Market data and methodology notes

• Daily split-adjusted prices were checked through September 1, 2026 against the public CSGP price-history download. The final close was also cross-checked against an independent market-data series.
• Historical CoStar ratios and the current peer screen were checked against public, S&P Global Market Intelligence-sourced pages for CSGP, Zillow, S&P Global, Moody’s, Verisk, FactSet and Gartner. Peer figures are secondary-source cross-checks, not substitutes for issuer filings.
• Five-year multiples and peer figures are period-end reference data reconstructed from public filings and market prices. They are rounded and used only for context; transaction timing, acquired revenue, noncontrolling interests and differing EBITDA definitions limit comparability.
• Management call statements are treated as management claims unless independently corroborated. Simple calculations in this report—growth, margins, annualized churn, acquisition multiples, static post-close net debt and implied perpetuity cash flow—are the author’s arithmetic from cited inputs.

Prepared from public information available through September 2, 2026. This report is general research, not personalized investment advice. Forecasts and scenario calculations are uncertain; readers should verify primary filings and consider their own objectives and risk tolerance.