CoStar Group Inc (NASDAQ: CSGP) — Margin Recovery Meets a Shrinking Growth Cushion

Published: 2026-10-28 · Verdict: Watch · Entry price: $25 · Research confidence: High (86%)

Executive conclusion

Analyst Take

WATCH, with medium-low investment conviction. CoStar has established a meaningful profitability recovery, but the evidence for renewed customer growth and disciplined capital deployment has weakened. My conditional entry reference is $25 per share, about 9% below the central cash-flow scenario of approximately $28. That is a modest valuation cushion, so operating confirmation remains essential. The comparable market reference is the verified October 27 regular-session close of $32.01, before the results and acquisition announcement. Reuters subsequently reported $28.30 after hours. Neither observation is an October 28 closing price; this update was researched before that session ended. [S10] [S29]

The strongest improvement is broader than an adjusted earnings presentation. Q3 operating income was $84 million, and Residential produced $16 million of company-defined EBITDA before the additional exclusions used in adjusted EBITDA. Residential’s adjusted EBITDA reached $49 million. The company has demonstrated that it can reduce the subsidy embedded in its portfolio. That materially weakens the claim that management cannot bring spending under control. It does not demonstrate that Homes.com itself is profitable: Apartments.com and acquired businesses contribute to the same segment, and management declined to disclose Homes’ quarterly loss. Both versions of EBITDA remain non-GAAP measures. [S1] [S2]

The disconfirming evidence is unusually clear. Annualized net new bookings fell to $37 million from $84 million a year earlier and $69 million in Q2. Apartments.com’s paid-property count declined sequentially, Homes membership was approximately unchanged, and Domain’s comparable revenue fell. The reported revenue-guidance increase includes about $60 million from Zonda. Removing that contribution reveals a $35 million reduction in the continuing-business outlook. Moreover, Q3 revenue excluding Zonda missed the previous company guidance midpoint even though consolidated revenue exceeded it. Expense control improved faster than retained customer acquisition. [S2] [S5]

The proposed $730 million Vexcel purchase intensifies the capital-allocation question. Its imagery and recurring data contracts fit CoStar’s established information strategy, and the acquisition offers a more tangible proprietary-data rationale than another effort to purchase consumer attention. However, disclosed trailing revenue and margin imply only about $28 million of adjusted EBITDA at the rounded starting point, roughly a 26-times acquisition multiple. The purchase would raise static net funded debt from approximately $500 million to $1.23 billion before intervening cash generation and closing adjustments. Strategic fit and adjusted-EPS accretion cannot establish an adequate return on that capital. [S6]

My constructive variant is that the market may underestimate the cash that can emerge from established professional information, Apartments.com, campus completion, and a smaller Homes loss. CoStar does not need a vigorous property-sales cycle for lenders and owners to require reliable information. AI could improve the productivity of its maintained database, while additional modules can use existing customer relationships. The strongest counter-case is that management removed productive sales capacity, improved immediate margins, and will now reinvest the apparent savings in sales rebuilding and acquisitions. In that outcome, reported earnings rise before shareholders receive a durable increase in economic cash.

The rebuilt acquisition-inclusive DCF produces approximately $6, $28, and $52 per share across explicit stress, central, and strong-execution assumptions. These are present-value scenarios, not formal price targets. The central case requires substantial margin expansion and approximately $925 million of economic free cash flow to the firm in 2030, after a replacement-cost charge for stock compensation. About 82% of its enterprise value comes from terminal cash flows. The verified regular-session reference embeds more terminal cash generation than that central path.

Confidence in reported financial facts is high; confidence in long-run portal margins and acquisition returns is substantially lower. The immediate decision sequence is Q4 bookings, a second positive Residential EBITDA quarter, the 2027 savings-and-hiring budget, and the Vexcel financing and closing bridge. Sustained organic growth above 8%, recovering paid properties, disclosed Homes renewal cohorts, and capital returns covered by recurring cash would improve the judgment. Another constant-perimeter revenue reduction, renewed Residential losses, or additional major acquisitions without return disclosure would weaken it. A lower quotation alone would not resolve those operating questions.

Changes since 2026-09-02

The September thesis passed one operating test and failed one capital-allocation test. Residential crossed company-defined EBITDA breakeven in Q3, although the requirement for consecutive positive quarters remains open. The commercial franchise retained strong renewal and margins, while the concern that cost reductions could impair bookings gained support. The approximate post-Zonda net-debt estimate was confirmed by September’s reported balance sheet. Vexcel then broke the earlier condition that major acquisitions pause until existing integration returns become observable. [S1] [S2] [S6]

Two earlier assumptions require explicit correction. The Capital Allocation Committee terminated on January 27, 2026; the full board assumed oversight. It was already stale to describe that committee as an ongoing restraint in September. The five-year price series also places the interval’s intraday low at $25.894 on July 29, rather than treating July 23’s closing observation as the intraday minimum. The September entry range was reached later that month, which records a triggered condition rather than a successful investment outcome. [S7] [S11]

The current financial tables also correct Q1 2026 company-defined EBITDA to $81 million and Commercial adjusted EBITDA to $161 million. The prior figures of $89 million and $168 million should not enter trailing calculations. The regulatory discussion now distinguishes the FTC’s proposed stipulated Zillow–Redfin order from independently verified court entry. [S1] [S20] [S33]

Stock Price Action — Five-Year Event Map

Company Financials’ split-adjusted series contains 1,254 daily observations from October 28, 2021 through October 27, 2026. The first close was $87.57 and the latest completed-session close was $32.01, approximately 63% lower. The interval’s intraday high was $100.38 on March 18, 2024; its intraday low was $25.894 on July 29, 2026. The trailing 52-week range was $25.894–$78.95. The reference close therefore sat about 12% of the distance from that low to that high. Closing and intraday observations are deliberately distinguished. [S10] [S11]

• Late 2021–2022: The stock moved from the $87.57 starting close to $77.28 at year-end 2022. Higher required returns are a plausible valuation influence, while the filings show continued revenue and operating-profit growth. The price series does not isolate an interest-rate contribution. [S14] [S15]
• 2023: The year-end close rose to $87.39 as residential investment expanded. The interpretation is continued willingness to capitalize marketplace potential despite lower current operating margins. [S13]
• 2024: The shares reached the $100.38 March intraday high and ended December at $71.59. Homes commercialization and increased marketing and campus expenditure changed the cash profile. Their precise contribution to the rerating cannot be measured from event timing alone. [S12]
• 2025: The year ended at $67.24. Matterport and Domain enlarged the business and acquisition capital, while consolidated operating earnings remained weak. Acquired scale and disappointing cash capture were simultaneous facts. [S3]
• January–July 2026: The stock closed at $61.66 on January 7 and reached its interval low in July. Financial targets, repurchases, residential uncertainty, and emerging search competition accompanied the decline; a single-cause explanation is unsupported. [S9] [S19]
• September–October 2026: The close fell to $26.95 on September 29, recovered to $35.45 on October 22, and declined to $32.01 on October 27. That last session fell about 5% before the evening release. [S11]
• October 27 after hours: Reuters reported $28.30 following the results and Vexcel announcement. That separate trading observation shows an adverse response to the combined disclosures, without identifying how much came from bookings, guidance, or acquisition spending. [S29]

Verdict: the price history establishes a large expectations reset. It does not prove competitive share transfer, permanent impairment, or undervaluation. The relevant next comparison is between retained operating economics and the cash recovery now embedded in the equity. [S1] [S2]

Business Overview

Customer value and monetization

CoStar sells maintained information, software, consumer attention, and transaction access. The customer proposition is understandable: reduce the time, uncertainty, and coordination involved in finding, evaluating, financing, marketing, leasing, and managing property. The analytical difficulty lies in separating the economics of those activities. A lender buying portfolio information values dependable history; a landlord buying advertising values qualified renters; an agent buying Homes promotion values listing exposure and client acquisition. Their budgets, alternatives, renewal incentives, and sensitivity to a property cycle differ. A single subscription label does not make those relationships economically identical. [S3] [S4]

The flagship CoStar product aggregates property, tenant, lease, transaction, ownership, and market information into a professional workflow. Its value depends on the probability that a customer finds an accurate answer quickly enough to improve a decision. An incomplete comparable set can produce a costly underwriting error; a missing tenant record can waste prospecting time. Consistent coverage across colleagues and markets reduces reconciliation work. This makes reliability and maintained breadth more important than the marginal cost of displaying a datapoint. The database must continue to be researched and refreshed; historical collection expenditure is an asset only while the resulting information remains relevant.

LoopNet sells commercial-property visibility. Its revenue depends on advertisers believing that placement reaches prospective tenants and buyers. STR benchmarks hospitality performance, using contributed information and established definitions. Real Estate Manager and Visual Lease support lease administration and accounting. Ten-X depends more directly on transactions. Matterport combines spatial subscriptions with capture products and services. These businesses can share content, distribution, and technology, but a common owner does not give every product the retention or margin structure of professional information. Product-level forecasting should precede consolidation.

Apartments.com sells rental advertising and associated tools to property owners and managers while allowing renters to search. A large community has continuing vacancies and repeated leasing needs. A small property may leave after filling its few units, then return later. Homes.com sells agents recurring promotion under a listing-agent-oriented proposition. Domain includes property-specific advertising tied to Australian listings. Zonda adds builder information, software, and marketplaces. These models connect to the property stock through different mechanisms: monitoring, vacancy turnover, new listings, construction, and completed transactions.

Reporting boundaries

The current portfolios are Commercial Real Estate and Residential Real Estate. Q3 revenue was $492 million and $460 million, respectively. Commercial included $344 million from CoStar, $89 million from LoopNet, and $59 million from other commercial products. Management separately identified approximately $320 million for Apartments.com, $31 million for Homes.com, and $70 million for Domain. Zonda entered Residential after its August close. These disclosures locate the largest revenue engines but do not constitute standalone financial statements for each brand. [S1] [S2]

This architecture is critical to interpreting the profitability milestone. Residential combines a mature rental marketplace, developing Homes operations, international portals, and acquired earnings. Its positive EBITDA can arise from less marketing, lower Homes losses, Apartments profit, acquisition contributions, or shared-cost changes. The segment result is a useful funding milestone. It cannot establish Homes contribution margin, cumulative project return, or standalone cash breakeven. Management’s refusal to give the quarterly Homes loss leaves that distinction unresolved rather than disproving the segment improvement.

Stability and recognition

Revenue is predominantly recurring, but its contractual stability has declined as the perimeter changed. The June filing puts subscription-based revenue near 89% in Q2, versus 95% a year earlier. Management said annual-contract subscription revenue represented 72% of Q3 revenue, with an 88% contract renewal rate and 94% renewal among customers of at least five years. These statistics describe different populations and should not be combined into one retention rate. [S2] [S4]

Straight-line recognition gives the group time to repair sales execution. Existing contracts generate revenue after bookings weaken, and recent additions enter recognized revenue gradually. This creates useful visibility but also delays evidence of reduced acquisition productivity. Conversely, improved bookings will not immediately restore the reported growth rate. Annualized bookings measure changes in contractual subscription value, including cancellations; they are neither quarterly revenue nor an unconditional cash backlog. The current gap between revenue growth and bookings therefore deserves attention without being translated mechanically into an identical future revenue decline.

Recurrence also differs from customer affordability. A broker can reduce users, consolidate offices, or cease trading. An agent can reach an anniversary and decide promotion did not justify its cost. A landlord can cut advertising after occupancy improves. Stable renewal in an established base provides evidence of usefulness, but it does not guarantee growth or protect every product from cyclical budgets. The most resilient customer is one who repeatedly uses information in necessary decisions, rather than one who merely signed an automatic-renewal contract.

Assets and the shareholder claim

The most valuable assets not fully recognized on the balance sheet are verified historical information, research processes, customer relationships, professional workflows, internally developed software, and consumer brand familiarity. Much of their creation cost was expensed. Reproducing them requires more than buying computers: it requires coverage, quality control, trusted definitions, distribution, and a period in which customers adopt the product. Yet historical spending is not a minimum asset value. Stale data, ineffective advertising, and obsolete software can have little economic worth despite their original cost. [S3]

Acquired goodwill records purchase expectations, not independently established success. A company can own a valuable organically created database while earning inadequate returns on recent purchases. Tangible book value consequently understates some operating assets while offering little protection against overpayment for others. The relevant test is retained cash profit relative to all committed resources.

International expansion adds currency, local housing conditions, local listing arrangements, and entrenched consumer habits. Shared technology can reduce duplicated development, while local content and distribution must still be earned. CSGP is Delaware corporate common stock listed on Nasdaq, not an ADR, MLP, or partnership interest. Ordinary ownership does not create partnership K-1 reporting merely by holding the shares; individual shareholder taxation remains circumstance-dependent. [S3] [S4]

Verdict: the economic drivers are understandable, and professional recurrence supplies resilience. The portfolio’s growing mixture of information, vacancy advertising, property-specific listings, software, capture, and transaction services makes a uniform subscription valuation increasingly inadequate.

Industry Dynamics

Distinct markets and profit pools

CoStar participates in several related industries with different sources of surplus. Professional CRE information rewards accurate breadth, dependable history, and repeated workflow adoption. Portals reward the aggregation of consumer intent and its conversion into advertiser outcomes. Lease software rewards implementation reliability and integration. Spatial and aerial information reward capture economics, refresh frequency, processing, rights, and downstream applications. A large property market can support all these services without making their margins or entry barriers interchangeable. The size of the underlying asset class does not determine the revenue available to a particular supplier.

The relevant competitors follow those purchase decisions. Zillow is directly relevant to rental advertising and residential attention. Rightmove and REA are important local benchmarks for OnTheMarket and Domain. Moody’s, MSCI Real Assets, and Altus overlap with professional information, institutional transaction intelligence, risk, benchmarking, and valuation workflows. CBRE and JLL principally supply services and brokerage: they are useful indicators of customer activity, but poor direct valuation substitutes. The existence of sophisticated rival datasets also contradicts a sweeping claim that CoStar monopolizes all CRE information. [S16] [S21] [S22] [S23] [S24]

Established information businesses can earn high margins because the decision being supported is valuable relative to the subscription. Collection and normalization require fixed expenditure; distributing additional access costs relatively little. Established local portals can also be very profitable when consumer habit and advertiser demand reinforce each other. Rightmove reported a 69% underlying operating margin in H1 2026, alongside a lower 66% statutory margin. That demonstrates an attractive mature profit pool, while the difference between incumbent economics and challenger spending remains fundamental. [S21]

Demand, geography, and market size

There is no single independently verified addressable revenue pool that reliably captures this portfolio. Customer willingness to pay, reachable account numbers, local competitive prices, and the cost of producing an adequate outcome matter more than the total value of buildings. For professional information, useful growth can come from additional users, geographic coverage, modules, portfolio monitoring, and higher wallet share. For portals, listings, vacancy turnover, marketing budgets, consumer discovery channels, and advertiser conversion constrain monetization. Counting every licensed agent or every property as an equally valuable prospect would exaggerate economic opportunity.

Established U.S. franchises remain central, while Australia, the United Kingdom, and other international markets increasingly affect group results. Domain’s comparable decline is evidence that local exposure can offset U.S. subscription growth. Management attributes part of that weakness to Australian transaction conditions; the disclosed revenue alone cannot separate market contraction from relative competitive performance. Commercial international launches have a different adoption process from residential portals: institutional customers may value cross-border information before a local consumer brand achieves widespread habit. [S2] [S3]

A weak transaction market has mixed effects. It reduces brokerage income, agent budgets, and some transaction fees. It can simultaneously increase the need for refinancing analysis, lender monitoring, vacancy marketing, and distressed-property information. Those offsets help explain why a maintained database can remain useful during a difficult cycle. They do not make all property-related spending countercyclical. Budget pressure can eventually reduce seats, delay new modules, and push smaller customers out of business.

Entry barriers and the capital cycle

The central CRE information barrier is the cost of producing credible coverage before enough customers pay to support it. An entrant can create a focused dataset or a better interface, but broad geographic coverage and trusted history require sustained collection and adoption. Customer skepticism makes early losses difficult to avoid. These conditions restrict effective new supply even when incumbent profitability looks attractive. The moat should appear in renewal, monetization, and research productivity; if those deteriorate persistently, an impressive database count is insufficient protection.

Portal supply responds more readily to attractive incumbent margins. Listings can be syndicated, consumers can use several sites, and adjacent platforms can redirect existing audiences. Competitors can therefore invest heavily without reproducing every historical record or persuading users to abandon all other portals. CoStar itself represents new capital entering that contest. The resulting supply response can absorb much of the apparent growth opportunity through advertising, sales hiring, subsidized pricing, and distribution payments. Industry demand can expand while the return on incremental investment falls.

Competition is intensifying in rentals. Zillow’s Q2 Rentals revenue grew 31%, including 42% multifamily growth, while Apartments.com’s subsequent Q3 growth slowed to 6%. Different periods and definitions prevent a precise share calculation, but the divergence makes an economically irrelevant-rival assumption untenable. Management’s account of aggressive competitor incentives is consistent with a contest for advertising budgets, rather than proof that price discipline alone will produce the best outcome. [S2] [S16]

Regulation and technology

The FTC announced a Zillow–Redfin settlement in August and filed a proposed stipulated order. Its remedies contemplate an independent Redfin rental advertising business and customer mobility, while allowing continued syndication. The reviewed public documents do not establish the court-entry date, so the six-month rebuilding deadline must be measured from verified entry rather than assumed to start on announcement. More independent advertising supply could increase customer choice; it is not automatically an exclusive benefit for Apartments.com. [S20] [S33]

Data access, copyright, privacy, Fair Housing rules, and MLS relationships can influence both competitive access and compliance cost. Protection of photographs, contractual rights, and a database’s selection and arrangement is different from ownership of underlying facts. A forecast that relies entirely on preventing rivals from learning public property characteristics would overstate legal protection. Product reliability and customer utility must support continued payment.

Foreign low-cost labor is an indirect competitive threat because CoStar sells locally relevant information, software, and audience rather than a standardized manufactured product. Lower-cost engineering, remote research, and AI can nevertheless reduce competitors’ expenses. Google’s documented Home Listings Ads place for-sale property information and promoted agent contact inside U.S. mobile Search, using a HouseCanary-powered feed and lead-based pricing. This is a specific discovery-channel threat, not evidence of a complete substitute for every CoStar service. [S19]

Verdict: barriers remain strongest in verified professional information and established local consumer habit. The supply-side capital cycle is less favorable in portals, where well-funded rivals and distributors can contest incremental economics before incumbent revenue visibly declines.

Competitive Position

A bounded professional information advantage

CoStar’s strongest advantage combines collection scale, historical consistency, breadth, and embedded use. The customer avoids assembling several incomplete sources and reconciling their definitions. The supplier spreads maintained research across many accounts. The outcome should be high retention, sustained willingness to pay, and attractive contribution after the continuing cost of collection. Q3 CoStar product revenue grew 8%, subscriber accounts rose 17%, renewal was 92%, and Commercial adjusted EBITDA margin was about 37%. Those results support a functioning franchise through difficult property conditions. [S1] [S2]

They also identify its limits. Subscriber growth substantially exceeded revenue growth. Management attributes that gap to customer mix, which is plausible, but the disclosure does not isolate incremental paid seats, enterprise users, smaller accounts, or geographic pricing. User growth is evidence of reach before it is evidence of proportional monetization. A stronger conclusion requires retained revenue and contribution from the expanding base. Commercial adjusted EBITDA also declined slightly year over year, so the mature engine is not currently generating uniform operating leverage.

Switching costs are primarily operational: customers must replace records, retrain teams, reconcile definitions, change recurring processes, and tolerate possible coverage gaps. A national institution using information repeatedly faces more friction than an occasional individual user. Contracts add near-term inconvenience; long-term captivity depends on maintained relevance. Customers can reduce access, combine specialized suppliers, use internal research, or stop a workflow whose benefits no longer justify its price. A wide moat requires continued investment, which limits the amount of gross margin that can be treated as immediately distributable.

Moody’s markets credit and property data, MSCI supplies institutional transaction and market information, and ARGUS supports valuation and cash-flow forecasting. These services can coexist within one customer organization. Their presence does not disprove CoStar’s advantage in broker, owner, tenant, and property workflows. It does disprove the broader inference that a strong position in those workflows establishes control of every institutional CRE information function. Market-share statements require a defined product, geography, customer, and revenue denominator. [S22] [S23] [S24]

Contributed information and AI

Debt Solutions could deepen the advantage because contributed loan portfolios provide information that public transaction records cannot fully reproduce. Management describes a benchmark covering $1.5 trillion of CRE debt and more than 155,000 loans. Broader participation can improve the comparison available to each lender, while existing customer relationships reduce distribution friction. The mechanism is credible; the product’s scale, profitability, and incremental retention remain management disclosures rather than independently measured economics. [S2]

The proposed imagery acquisition could add physical observations to this information system. Combining building characteristics with ownership, tenants, leases, and use may be more valuable than either dataset separately. However, an aerial image cannot directly establish every interior, occupancy, contractual, or legal attribute. The inference layer must be validated, and customers must value its accuracy. More extracted fields can increase coverage while also increasing error propagation and remediation costs.

Management’s tenant-research AI claims should therefore be judged by cost per verified update, error rates across representative records, customer usage, and retention. An internal comparison against an AI model using public information is not an independent accuracy study. A tenfold productivity aspiration is not a tenfold realized margin benefit. Competitive tools may also improve, allowing customers to capture some savings through lower prices. The best outcome would be better information and retained pricing with slower research-cost growth; that combination has not yet been disclosed comprehensively. [S2] [S6]

Apartments.com: relevance without complete captivity

Brand matters economically when it attracts high-intent users, reduces paid acquisition, or improves advertiser conversion. Apartments.com has an established identity and substantial inventory. Management reports stronger visitor activity, tour requests, and applications, supporting continued relevance. Yet advertisers ultimately pay for incremental leasing outcomes, not unaided awareness or aggregate engagement. A traffic metric becomes more persuasive when linked to retained property spending and contribution after the cost of acquiring that traffic. [S2]

The adverse operating evidence is concrete. Paid properties ended Q3 at 91,155, down approximately 2% from June, and monthly renewal was 98.4%. Management points to insufficient account coverage and a greater mix of smaller properties. Those explanations can be valid while competition also matters. A small landlord who fills a vacancy may rationally cancel without being dissatisfied; a large community may retain several portals and allocate budgets dynamically. High aggregate renewal does not establish exclusive dependence.

Stable average revenue per property is consequently useful but incomplete. Holding price can protect customer quality and long-run positioning. It can also sacrifice volume if customers find adequate alternatives. Falling average yield can indicate discounting, or a sensible move into smaller accounts. Matched property-size cohorts, same-property spending, reactivation, and contribution after selling costs would distinguish those mechanisms better than the headline average.

The attribution challenge is a useful diagnostic project. Management says initial customers discovered more Apartments-originated leases than their existing reporting recognized. However, first-touch attribution is not incremental causality: a renter might have leased the property through another channel anyway. Early participants may also differ from the broader advertiser base. The study can improve measurement and sales conversations before it establishes a representative financial return.

Homes and international challengers

Homes’ listing-agent proposition can appeal to professionals who want inquiries to preserve their own customer relationship. That is a credible product distinction, but agents can advertise across several channels and consumers can switch interfaces cheaply. Durable advantage requires retained customers whose incremental profit supports pricing and acquisition cost. Q3 membership remained above 36,000 and approximately flat sequentially. Platinum’s September launch and higher October new-sales pricing show monetization potential, not an established cohort return. [S2]

Premium placement creates a further test: can additional retained gross profit exceed retargeting, photography, support, and selling expense without merely moving spending from the basic package? Agent productivity can create selection bias in customer-success comparisons. High demo conversion also excludes prospects who never accept a demonstration. Neither metric alone supplies lifetime value.

Domain and OnTheMarket retain local brands and customers, while REA and Rightmove remain the relevant incumbent standards. Shared technology may lower duplicated development, but local listing rights, advertiser relationships, consumer habit, and distribution must still be won. Verdict: the professional information moat is supported by observable retention and margins. Apartments has meaningful audience advantages with incomplete customer captivity; Homes and international portals still need retained economic proof. [S2] [S21]

Growth History and Forward Opportunities

Growth sources matter more than consolidated size

Revenue rose from approximately $1.94 billion in 2021 to $3.25 billion in 2025, about 14% annually. Operating income moved from $432 million to a $72 million loss. Growth was therefore accompanied by a major change in spending intensity and acquisition capital. The record supports the ability to expand revenue, but it does not establish that each incremental dollar earned an adequate shareholder return. The portfolio’s present challenge is to turn a larger gross-profit pool into retained cash without sacrificing the sales engine. [S3] [S14] [S15]

Q3 reported revenue growth was 14%, while management identified 7% organic growth. Residential grew 23% reported and 8% organically. Acquisition annualization, new purchases, and currency therefore belong in separate forecast bridges. Purchased revenue can be valuable, but it should not be described as evidence that existing products accelerated. Conversely, organic growth is not automatically high-return growth if it requires disproportionately expensive acquisition or service capacity. [S2]

Professional products have the clearest distribution advantage

Debt benchmarking, tenant information, lease administration, and commercial geographic expansion can use existing professional relationships. A lender already relying on property records may adopt another underwriting tool without requiring a new consumer brand to be created. Shared research and software can spread fixed costs. This is a more direct economic mechanism than treating every new product as another large addressable market. Success still requires a useful workflow, disciplined complexity, and selling effort that produces incremental retained contribution.

Timing matters. Management expects meaningful Australian commercial bookings primarily in the first half of 2027, with a combined Real Estate Manager and Visual Lease opportunity later in 2027. Vexcel-related enhancements are also more intermediate-term than immediate. Debt benchmarking is nearer term. Product availability, sales readiness, initial bookings, recognized revenue, and cash return are separate milestones. Announcing all of them together does not justify putting every benefit into the next quarter’s forecast. [S2]

LoopNet offers encouraging evidence: revenue grew 13%, and its larger salesforce accompanied stronger bookings. Coordinated selling to smaller CoStar and LoopNet customers may reduce duplication and improve coverage. The U.K. trial supports further rollout, but it is not a controlled estimate of the U.S. result. Better local demand or easier opportunities could partly explain the stronger performance. The test is incremental net revenue and contribution after the complete selling cost.

Rentals and Homes need retained account growth

Apartments.com’s immediate opportunity is to convert existing audience relevance into renewed paid-property additions. Management plans more field selling because servicing and prospecting capacity lagged growth. If correct, bookings should improve before revenue fully responds. New representatives require recruiting, training, salary, supervision, and ramp time. Historical gross production among established representatives cannot be assigned immediately to each new hire. Same-property revenue and retained contribution will determine whether hiring fixes a capacity constraint or merely increases spending in a more competitive market. [S2]

Homes has price, premium mix, and penetration opportunities. However, its approximately flat subscriber base indicates a replacement challenge. The last comparable June cancellation disclosure was 2.4% monthly; sustained uniformly, that would imply roughly 25% annual attrition. This is arithmetic sensitivity, not a cohort forecast. July’s June subscriber price of $305, September’s $271 subscriber price, October’s $429 new-sales price, and recognized revenue are not one consistent ARPU series. Their definitions must be reconciled before they support a precise lifetime-value model. [S2] [S5]

The product outlook is strongest where new functionality can use an existing professional relationship; the near-term portal outlook depends on restoring retained customer growth after the cost reset. Premium sales can increase revenue with unchanged membership, but continued acquisition spending requires evidence that those higher-priced relationships survive anniversaries and earn contribution after service delivery.

Acquired growth and integration

Matterport subscription revenue grew 11%, and management expects profitability by the end of 2027. Domain’s proposed common platform with Homes could reduce duplicated technology expense, while current transaction conditions remain difficult. Zonda adds builder information and marketplaces; Vexcel would add another recurring-data platform. These are credible product rationales rather than proof of purchase returns. A common platform can save cost while creating migration, customer-support, and development risks that delay monetization. [S2] [S6] [S8]

Zonda contributes an estimated $60 million of 2026 revenue and $10 million of adjusted EBITDA under current guidance. Full-year ownership adds a comparison benefit in 2027; Vexcel is excluded from the current outlook. Forecasts should separately identify acquisition annualization, underlying growth, integration costs, and synergies. Verdict: professional extensions offer the clearest route to attractive incremental economics, while portal recovery and acquired growth require a more explicit bridge from adoption to retained cash. [S2]

Financial Quality

Five-year reconstruction

The following filing-based figures are rounded U.S. dollars in millions. Conventional free cash flow is operating cash flow less purchases of property, equipment, and other assets, including campus expenditure, before acquisitions. Period-end shares differ from weighted-average earnings denominators. Company Financials’ statements were used as a cross-check; primary statements control classification and material values. [S3] [S12] [S13] [S14] [S15]

The business expanded its gross-profit pool while expenditure grew faster. This was substantially a management-directed investment cycle, with campus development and consumer marketing changing the conversion profile. Nevertheless, approximately $933 million of cumulative conventional FCF over five years is the cash outcome shareholders received before acquisitions and capital returns. Calling expenditure investment does not establish its return.

Earnings are recovering from an internally depressed investment period, rather than a demonstrated cyclical peak. Property markets remain difficult, but the scale of consolidated margin compression was heavily influenced by spending decisions. Recovery can therefore occur without a transaction boom. Its durability depends on whether removed expenses were ineffective or necessary to sustain future demand.

Current operating bridge

These current issuer tables correct the earlier Q1 reconstruction. Trailing revenue is approximately $3.674 billion, company-defined EBITDA $546 million, and adjusted EBITDA $722 million. Q3 gross margin was about 77.8%, while operating margin reached 8.8%. Operating improvement therefore came principally below gross profit, through expense control, rather than an expanding gross margin. [S1]

Residential supplied more than the entire year-over-year increase in consolidated adjusted EBITDA; Commercial declined by $3 million. That does not make the commercial franchise weak, but it concentrates the incremental profit bridge in the segment with the least product-level profitability disclosure. The recast annual record likewise separates Commercial adjusted EBITDA of $536 million, $602 million, and $672 million in 2023–25 from Residential results of negative $44 million, negative $361 million, and negative $230 million. Mature profit and development subsidy coexist. [S3]

All EBITDA measures are non-GAAP. Q3 adjusted EBITDA exceeded company-defined EBITDA by $50 million, including $32 million of stock compensation. In 2025, Company Financials’ mechanical operating-income-plus-cash-flow-D&A calculation gives $191 million, versus the issuer’s $170 million EBITDA. The filing reconciles the $21 million difference to lessor depreciation and amortization reported outside operating income. The analogous 2024 difference was $29 million. Numerator definitions must be reconciled before comparing multiples. [S1] [S3] [S4]

Cash, compensation, and accounting

Nine-month operating cash flow was $455 million; campus purchases were $134 million and other capital purchases $60 million, producing $261 million conventional FCF. Subtracting $112 million of stock compensation gives an illustrative $149 million economic-cost sensitivity. Adding back only the $109 million Brown judgment payment gives $370 million before that compensation charge and $258 million after it. The payment is unusual for operating comparison but remains part of acquisition economics. [S1] [S4]

Trailing operating cash flow is approximately $617 million and capex $260 million, leaving $357 million FCF. Adding Brown produces $466 million; subtracting approximately $153 million of trailing compensation produces about $313 million. This compensation deduction is a replacement-cost convention, not cash actually paid. An alternative model may project dilution and offsetting repurchases. Charging full compensation, equivalent dilution, and all offsetting repurchases simultaneously would double count ownership cost.

Operating cash exceeds net income because noncash depreciation, amortization, and compensation are added back, while commissions, working capital, taxes, and legal payments affect cash separately. Nine-month depreciation and amortization of $265 million and compensation of $112 million are material beside $132 million net income. Deferred commission amortization was $93 million against $97 million of new deferred commissions. Cash conversion is understandable, but not captured by a single adjusted margin. [S1]

Accounting is conventional rather than demonstrably unusually conservative. Research and advertising are generally expensed, while qualifying contract-acquisition commissions are deferred over a three-year benefit period. Purchase accounting recognizes estimated customer relationships, technology, brands, and goodwill. Expensed internal investment can understate economic assets while acquired capital and selling costs remain recognized. Neither observation justifies treating every noncash expense as costless. [S3] [S4]

Return on capital

Estimated acquisition-inclusive ROIC uses operating income multiplied by an assumed 75% after-tax factor, divided by average equity including NCI plus $1 billion funded debt principal less unrestricted cash. Goodwill remains included. Leases are treated after rent. This is a reproducible funded-capital proxy, not a disclosed company metric or a measured return on the mature database. Corporate property held partly for leasing also remains in capital while related results appear outside operating income, limiting exact operating matching. [S3] [S12] [S14]

Under that convention, estimated ROIC falls from 11.7% in 2022 to 7.0% in 2023, 0.1% in 2024, and -0.9% in 2025. Average 2025 capital is $5.805 billion. A matched three-year research sensitivity adds current development plus two-thirds of the previous year and one-third of the preceding year to ending capital. It replaces current expense with amortization of the three preceding cohorts.

The resulting 2025 research asset is about $713 million, versus $578 million in 2024. Replacing $406 million development expense with approximately $272 million amortization raises operating income to about $62 million and ROIC to roughly 0.7%. The numerator effect dominates the larger denominator. Research capitalization therefore raises this loss-stage return, contrary to a mechanical transfer from profitable software examples. Useful life, obsolescence, acquired-technology overlap, and tax treatment remain assumptions.

A favorable sensitivity retaining compensation but excluding acquired amortization and specified adjustments produces approximately $198 million of operating earnings and a return near 2.6%. That is not GAAP ROIC and excludes potentially recurring costs. Even this treatment remains below an assumed 8–10% capital hurdle. The evidence supports weak demonstrated consolidated returns, without establishing the undisclosed standalone return of the commercial franchise.

Capital intensity and obligations

Physical capital intensity was temporarily elevated: campus purchases were $579 million in 2024 and $307 million in 2025. Completion should improve FCF mechanically. However, the June filing also records $10 million of H1 interest capitalized into construction. As construction ends, more interest can reach earnings even while capital spending falls. Imagery capture adds continuing equipment and refresh requirements. [S3] [S4]

At September, cash was $500 million, debt carrying value $994 million, and total equity $7.992 billion. Goodwill plus net intangibles totaled $7.490 billion. The $1 billion notes bear 2.8% interest and mature in 2030; the $1.1 billion revolver matures in 2029 and was described as undrawn on the call. Funding capacity is meaningful, but is not unrestricted surplus after operating needs and Vexcel. [S1] [S2] [S3]

Material economic obligations include leases, purchase commitments, construction, legal claims, and prospective acquisition payments. The December filing disclosed $205 million purchase obligations and $155 million remaining construction commitments; those dated amounts should not be treated as unchanged September liabilities. Brown’s payment does not resolve separately disclosed similar Matterport claims. Vexcel is a conditional future commitment rather than existing September debt. Verdict: cash and earnings are recovering, but historical capital returns remain weak and normalized cash must be separated from capex completion, compensation exclusions, and acquisition deployment. [S3] [S4] [S6]

Capital Allocation

Deployment still exceeds internally generated cash

Nine-month conventional FCF of $261 million compares with $668 million of program repurchases and $789 million of acquisition cash payments net of acquired cash. Combined deployment exceeded FCF by approximately $1.20 billion before other financing movements. Existing liquidity funded the difference. The policy is reinvestment and portfolio expansion with repurchases, rather than harvesting a mature information franchise primarily for distribution. [S1]

Reinvestment in research density, contributed information, and additional professional workflows has a plausible route to attractive incremental returns. It can improve products already embedded in customer decisions. Simultaneously funding consumer audience acquisition, international portals, spatial capture, builder data, and aerial imagery creates a different portfolio risk. Each project can have strategic logic while the combined program exceeds management’s integration capacity or the cash available for shareholder returns.

Acquisition scorecard

The Matterport and Domain purchase-accounting bridges should replace simplified headline enterprise values. Acquired cash, investments, assumed obligations, and post-combination compensation require separate treatment. Matterport also illustrates why normalizing a legal payment out of one period’s operating cash does not erase its economic acquisition cost. A profitable acquired product can still fail to earn an adequate return on the purchase capital. [S3] [S4]

Management reports that Matterport is improving and expects profitability by late 2027. That is a useful operating milestone, not a return calculation. Domain’s common platform could reduce duplicated development, while its current comparable decline emphasizes local cyclicality. Investors need post-acquisition cash contributions, customer retention, and integration expenditure against full committed capital, rather than only evidence that each product has grown or remains strategically relevant. [S2]

Zonda’s $800 million closing announcement, the call’s $815 million description, and $789 million net acquisition cash flow use different possible bases. Closing adjustments and acquired cash may reconcile them, but an invented bridge would be false precision. At the announcement price and disclosed 2025 earnings, the starting multiple is about 20.5 times adjusted EBITDA. The final allocation and transaction-level cash reconciliation remain necessary. [S1] [S2] [S8]

Vexcel’s rounded starting adjusted-EBITDA yield is under 4% before tax, capex, and financing. High trailing growth and data synergies may improve that return materially. Management compares adjusted-EPS accretion favorably with repurchases, but financing assumptions and excluded amortization can make EPS rise before an investment clears its cost of capital. The board should disclose a cash-return hurdle and subsequent scorecard, including the capture infrastructure required to maintain the dataset. [S2] [S6]

Repurchases, issuance, and personal capital

The company repurchased approximately 16.5 million shares for $668 million in the first nine months, about $40.50 per share. Q3 purchases were 2.7 million shares for $81 million, near $30 each. Later purchases were cheaper than earlier ones, while the year-to-date average exceeded the October 27 close. That hindsight comparison describes execution price, not intrinsic value at each purchase date. A repurchase creates value when the retired claim is worth more than its cost and the remaining company retains adequate funding. [S1] [S2]

Net share retirement is meaningful in 2026. The previous year’s $500 million program did not prevent ending shares rising from 409.5 million to 417.9 million, because Matterport issuance and employee awards offset purchases. The latest verified July cover-page count was 405.198 million; Q3 weighted-average diluted shares were 401.9 million. The latter supports a lower earnings denominator but is not an exact September or October ending count. [S1] [S3] [S4]

Material stock is issued to employees, directors, and executives. Compensation expense was $194 million in 2025, falling to $112 million in the first nine months of 2026. The reduction helps economic conversion, but repurchases do not make compensation free. Awards, option exercises, tax withholding, and discretionary purchases answer different questions and should not be combined into one insider-confidence statistic.

Florance’s verified open-market purchases include 55,720 shares at $44.52 in February, 71,430 shares across two May purchases, and 83,300 shares at $29.89 in August. The August purchase was approximately $2.49 million and was not marked as a prearranged trading-plan transaction. These personal commitments support alignment. They do not establish valuation accuracy or validate much larger corporate investments. [S25] [S31] [S32]

Incentives and oversight

The Capital Allocation Committee automatically terminated January 27, 2026; oversight moved to the full board. Its existence cannot be cited as a continuing independent restraint on Vexcel. The revised director program includes a $75,000 annual cash retainer, additional role retainers, and a $250,000 annual restricted-stock award. Equity creates exposure to share value, while strategic independence must be judged through actual decisions. [S7]

The 2026 CEO cash incentive uses 50% organic revenue and 50% EBITDA. Other named executives use 40% organic revenue, 40% EBITDA, and 20% individual performance. Performance-stock goals use 75% adjusted EPS and 25% relative shareholder return. Organic revenue counterbalances purchased scale, but adjusted EPS can benefit from repurchases and acquisition adjustments without demonstrating acquisition-inclusive ROIC. The earlier precise bespoke payout figures are not reproduced because their numerical bridge was not independently established. [S7]

The CEO’s legacy tax gross-up was removed, and the equity design eliminated new options and the prior restricted-stock vehicle. These are substantive improvements without a direct published return-on-capital incentive. CoStar has never paid a cash dividend and does not anticipate one; allocation coverage therefore concerns reinvestment and repurchases rather than dividend maintenance. Verdict: insider commitment and incentive improvements are real, while deployment above FCF and limited acquisition-return disclosure keep capital discipline an open economic test. [S3] [S7] [S9]

Changes and Headwinds — Last Two Years

The profitability harvest still includes expansion

The largest strategic change is a move from aggressive consumer investment and portfolio construction toward a stated profitability harvest, while acquisitions continue. Marketing and campus expenditure depressed 2024 cash generation. Matterport and Domain enlarged the business in 2025. Expense reductions and repurchases became central to the 2026 narrative, followed by Zonda’s closing and the Vexcel agreement. The company is releasing cash from some investments while committing resources to others; that is a continuing allocation cycle rather than a completed transition to distribution. [S3] [S6] [S8]

Internal actions explain much of the current profit improvement; external conditions affect transactions and advertiser budgets. Lower marketing, organizational changes, procurement, and integration influence costs directly. Australian listings, U.S. agent income, rental concessions, and CRE activity influence customer spending. Management’s macro explanations should be treated as hypotheses about contribution, not complete decompositions. The numbers can improve internally while underlying customer markets remain difficult.

Guidance on a constant perimeter

Management explicitly confirmed this bridge. Q3 revenue excluding Zonda was $933 million, approximately $7 million below the earlier guidance midpoint. Excluding Zonda’s $3 million adjusted EBITDA, profit exceeded the old guidance ceiling by $16 million. Thus the same quarter missed the prior revenue midpoint and surpassed the profit range. The contrast describes cost execution alongside weaker demand or sales conversion. [S2] [S5]

The Q4 outlook requires approximately $986 million revenue and $285 million adjusted EBITDA at the midpoints, about a 29% margin. It also places a substantial portion of the full-year Residential profit in Q4. Seasonal patterns and declining expense can explain part of that concentration, but another quarter is needed to establish durability. The unchanged $1.25 billion 2028 adjusted-EBITDA target does not incorporate a fully disclosed allocation of newly identified savings. [S1] [S34]

The sales diagnosis changed

Management describes approximately 2,000 sales professionals, down about 10% year over year, and a Homes inside-sales team of 267 versus a prior peak of 655. Recruiting is being decentralized, and field experience receives greater emphasis. The July narrative emphasized efficiency; October acknowledges insufficient coverage. These accounts can coexist if ineffective inside capacity was removed faster than productive field capacity could be built. They nevertheless imply a cost and timing gap before reacceleration. [S2] [S5]

The CFO separates roughly $150 million of 2026 savings into approximately $100 million expected to persist and $50 million mainly related to lower-than-anticipated sales headcount. The latter should not be capitalized as permanently available profit if hiring must resume. A further $200 million of opportunities is divided among organizational efficiency, integration, and procurement or marketing effectiveness. Management has not completed the budget allocating those savings between margins and investment. Announced savings are therefore neither fully incremental earnings guidance nor cash already available for distribution. [S2]

Management, facilities, regulation, and accounting

Robin Rossmann became CFO effective July 31, replacing Christian Lown. His operating background is relevant to an efficiency review across acquired and international businesses. Management’s claim that European experience can scale globally remains a forecast. A leadership transition during multiple integrations adds execution demands without establishing that turnover reflects a disagreement or accounting problem. [S26]

Campus expenditure is declining but remains material. Falling physical investment can release cash, while more depreciation and previously capitalized interest affect earnings. Vexcel would introduce additional capture equipment, processing, and recurring refresh expenditure. Management describes its current capex as approximately $8–10 million annually, with possible additional investment for CoStar applications. That is an initial estimate, not a permanent cap on reinvestment. [S2] [S4]

Brown was paid in Q2, while similar Matterport-related claims remain separate. CoStar’s copyright dispute with Zillow is also distinct from the FTC’s rental agreement settlement. Potential recoveries should not be treated as certain cash. The FTC’s reviewed announcement describes a proposed order requiring court approval; neither final entry nor Redfin’s completed rebuilding is established here. [S4] [S20] [S33]

Reporting changed from geographic to product-portfolio segments, with comparative periods recast. That changes visibility rather than organic growth. The company adopted the receivable expected-credit-loss practical expedient prospectively from January 1, 2026 and reported no material adoption impact. The reviewed evidence does not identify a material revenue-recognition change driving the profit recovery. Verdict: cost execution improved, but sales rebuilding, acquisition integration, search competition, and uncertain savings allocation make the permanence of the improvement the central next test. [S3] [S4] [S19]

Risk Analysis

The most plausible adverse outcome is disappointing shareholder cash returns rather than immediate financial distress. A useful information franchise can still be a poor investment if cash is repeatedly committed below the required return or if the share price capitalizes an overly favorable terminal margin. Risks interact: weak bookings can make sales rebuilding more expensive, while acquisitions reduce liquidity and increase the cash improvement required to support valuation. The following likelihood assessments are qualitative analyst judgments, not estimated probabilities. [S1] [S2] [S6]

Another stock decline could follow a further guidance reset, weak bookings after hiring resumes, renewed Residential losses, deteriorating rental retention, or a larger financing requirement. These are observable triggers, but they should not be treated as independent events. A single acquisition that disappoints can reduce profit, absorb cash, consume management time, and increase the required return investors apply to the entire portfolio. That interaction creates downside beyond a simple miss in one quarterly EPS estimate.

Operating underinvestment and cohort risk

The most immediate operating danger is an expense reduction that improves current earnings while weakening the future customer base. Subscription recognition can delay the revenue effect, making a few quarters look better before the damage becomes obvious. The alternative is genuinely productive simplification: redundant roles and ineffective marketing may disappear without reducing customer outcomes. The distinction requires bookings and retained contribution after the replacement sales cost, not a presumption that either all cuts were beneficial or all were harmful. [S2]

Homes has additional anniversary risk. New-sales pricing can rise faster than recognized revenue and may not survive renewal. Premium subscribers may be unusually productive agents, making aggregate success comparisons unreliable. A flat member count can hide both strong gross acquisition with heavy churn and weak acquisition with excellent retention. Neither supports a confident lifetime-value estimate without the missing cohort bridge. Attractive product engagement can coexist with poor fully loaded payback.

Platform and acquisition dependencies

The AI risk runs in two directions. Better extraction can reduce CoStar’s research cost; better generic discovery can weaken the monetization of a portal front end. These can occur simultaneously across different products. Distribution concentration matters even without a collapse in visits: the highest-value intent could shift to paid surfaces, increasing the cost per retained advertiser. Measuring aggregate traffic without channel contribution may therefore miss economic leakage. [S19]

Vexcel also brings concentration risk: the issuer specifically warns of dependence on a limited number of large customers. Large technology licensees may be valuable endorsements while possessing bargaining power, internal alternatives, and different strategic interests from property advertisers. Aerial capture requires refresh, equipment, operational reliability, and rights. A rapid growth rate before acquisition does not establish that its largest contracts renew indefinitely at unchanged economics. [S6]

Severe and total-loss paths

A catastrophic investment loss could arise from major impairment of the commercial engine combined with continued cash destruction in adjacencies, a legal or data-rights shock, or financing stress. A prolonged security failure, loss of critical feeds, or inability to maintain information quality could impair the very reliability customers purchase. Shareholders need not encounter insolvency to suffer a severe permanent loss; sharply lower terminal cash expectations are enough when terminal value dominates the model. [S3]

Total loss is a low-probability, unquantified tail. A plausible path would combine a major operating or legal impairment, continued debt-funded commitments, prolonged negative cash generation, and an eventual inability to refinance or raise equity on survivable terms. Current recurring revenue, positive operating cash, and unused credit capacity argue against that as the central outcome. They do not eliminate the need to monitor discretionary commitments and the quality of the commercial cash source. [S1] [S2] [S4]

Verdict: the largest practical risk is the combination of weak retained growth and repeated reinvestment, rather than leverage alone. The established franchise provides repair capacity, but cannot protect shareholders from paying too much for future cash or allocating that cash poorly.

Valuation Discussion

Capitalization and comparable denominators

Using the verified $32.01 October 27 close and July’s identifiable 405.198 million ending shares produces approximately $12.97 billion equity value. Adding $1 billion debt principal and $17 million NCI, then subtracting $500 million cash, produces approximately $13.49 billion enterprise value. This is a reference capitalization with a dated share denominator, not an assertion that July shares remained unchanged. Using 400 million shares instead reduces equity value by about $166 million, or 1.3%. Exact ending shares should be refreshed when disclosed. [S1] [S4] [S10]

Vexcel has not closed. A static acquisition-inclusive commitment adds $730 million, taking the comparable enterprise commitment to approximately $14.22 billion and net funded debt to $1.23 billion. These estimates exclude intervening cash, financing fees, acquired cash or debt, working-capital adjustments, and the operating liquidity retained. Acquired operations must be included in the cash forecast if their purchase commitment is included in the equity bridge. [S6]

These measures answer different questions. FCF from operating cash is after cash interest, so equity value is the appropriate numerator for those yields. The DCF below uses unlevered cash and enterprise value. Operating leases remain under an after-rent convention; adding their liabilities without matching the earnings treatment would distort the comparison. Forecast compensation of $152 million is material to the apparent forward multiple. [S1] [S3]

Own history and peers

Company Financials’ period-end EV/sales series falls from about 14.7 times in 2021 to 12.6 in 2022, 12.9 in 2023, 9.4 in 2024, and 8.5 in 2025. Current compression is substantial, but historical values reflected different margins, cash balances, portfolio composition, and reinvestment expectations. Multiple mean reversion is therefore an outcome requiring economic justification, not a valuation method. [S11]

Zillow is the closest U.S. portal cross-check. Its October 27 close of $32.05 and June ending shares of approximately 224.9 million imply $7.21 billion equity value. Cash and investments were $682 million. Treating substantially matched mortgage warehouse borrowing and inventory as operating items gives operating enterprise value around $6.5 billion; including warehouse borrowing without a matched asset adjustment moves it toward $7.0 billion. Against its $2.92–2.96 billion revenue and $730–760 million adjusted-EBITDA outlook, the rough ranges are 2.2–2.4 times sales and 8.8–9.6 times adjusted EBITDA. [S11] [S17] [S18]

CoStar’s premium therefore needs its commercial franchise and better eventual economic conversion. Zillow’s larger compensation burden and different mortgage risk limit direct comparability. A matched warehouse-flow adjustment must normalize both financing and loan movements; removing borrowing alone would create a misleading underlying-cash signal.

MSCI illustrates information economics rather than a direct consolidated peer. At $601.75, 71.5 million shares, $6.76 billion debt principal, and approximately $257 million unrestricted cash, enterprise value is about $49.5 billion. Annualized Q3 revenue and adjusted EBITDA imply approximately 13.9 and 22.3 times, respectively. Its 62.2% adjusted margin and 56.3% operating margin explain why a large sales premium need not imply an equally large earnings premium. Annualization is a run-rate check, not a forecast. Rightmove’s 69% underlying margin supplies another incumbent benchmark, not a margin entitlement for challengers. [S11] [S21] [S27]

Explicit operating and cash scenarios

All following figures are analyst assumptions, not guidance. The model includes full Zonda ownership and assumes Vexcel closes, with no further major acquisitions. Compensation is charged as a cash-equivalent replacement cost and shares are held at 405.198 million. This avoids simultaneously charging compensation and equivalent dilution. Additional uncompensated dilution would reduce per-share outcomes; no separate value is credited for speculative repurchases.

FCFF equals adjusted EBITDA less compensation, less 25% tax on earnings after ordinary depreciation, less capex and incremental working capital. Ordinary depreciation assumptions for 2027–30 are $100/$105/$110/$110 million in bear and base, and $105/$110/$120/$125 million in bull. Acquired amortization is excluded from normalized operating economics, with no perpetual tax benefit assumed. For example, base 2030 is $1,650 million adjusted EBITDA minus $220 million compensation, $330 million tax, $140 million capex, and $35 million working capital: $925 million.

The base approximately reaches management’s $1.25 billion 2028 adjusted-EBITDA objective, then requires further expansion. The bear assumes limited organic growth and persistent acquisition costs; it does not require disappearance of the database. The bull needs productive hiring, premium adoption, modules, integration, and restrained reinvestment simultaneously. These operating assumptions connect the valuation to the disclosed growth and allocation questions. [S2] [S6]

The rates are analyst hurdle assumptions, not estimates established by the factor model. Annual discount periods approximate a year-end 2026 valuation. The equity bridge subtracts $1.247 billion of static post-Vexcel net debt and NCI. Required cash reserves, closing adjustments, timing, or non-operating property value can change the result.

Holding base interim cash and terminal assumptions constant, the regular-session reference requires approximately $1.08 billion of 2030 economic FCFF, about 17% above the base. The market is correctly capitalizing real commercial earnings and achievable cost reductions; it also requires durable further conversion. A segment valuation may show additional potential, but shared infrastructure and management’s continuing reinvestment policy limit separability. Verdict: the central valuation issue is cash capture after the recovery, with terminal margins and capital allocation more consequential than small share-count refinements.

Variant Perception

Consensus and the live investor debate

A verified numerical sell-side consensus valuation was not available in the reviewed evidence, so no consensus target or rating distribution is invented. The earnings Q&A provides a more reliable map of the debate: whether demand exists but cannot be reached, whether cuts went too far, whether savings will reach margins, and whether Vexcel earns a superior cash return. These questions concern the connection between operating improvement and shareholder value, rather than simply whether reported EPS exceeded expectations. [S2]

Peter Christiansen asked how to think about 2027 bookings reacceleration and margins. Curtis Nagle challenged the sales-capacity explanation and whether reductions were excessive. Surinder Thind asked why additional savings did not increase the 2028 target and how much would be reinvested. Craig Huber requested the Homes quarterly loss; management did not provide it. The most thoughtful investor questions target missing bridges rather than another traffic statistic.

Strongest opposing cases

The strongest bull case is that established information and rental earnings are hidden beneath inefficient expense and a shrinking development subsidy. Positive Residential EBITDA is evidence that the funding burden can fall. Lower campus expenditure can release cash without a housing boom, while contributed data and additional modules can improve professional customer value. Vexcel adds recurring observation infrastructure that might lower collection cost across the portfolio, rather than requiring another consumer audience to be built from scratch. [S1] [S2] [S6]

The strongest bear case is that sales capacity was removed faster than replacement productivity developed. Margins improved immediately, while bookings and paid inventory weakened. Rebuilding coverage could consume savings before revenue responds. Management may then direct the remaining cash into additional acquisitions whose returns are difficult to observe. The commercial moat can remain intact throughout that adverse shareholder outcome. Flat Homes membership and continued portfolio expansion support taking this possibility seriously.

Positioning and the factor model

The factor model’s October 27 snapshot reports Market exposure of 1.188, NewValue 0.842, SmallSize 0.705, and Liquidity 0.549. Real Estate and Financials statistical exposures are 0.508 and 0.397; Quality is -0.190 and Momentum -0.029. These are internal estimated return exposures, not industry classifications, balance-sheet facts, or fundamental quality ratings. Positive dividend-related exposure does not establish a dividend policy. [S28]

The factor model explains 34.0% of return variation, with adjusted R-squared of 31.8%. ResidualMomentum is -0.116 and ResidualSharpe -1.321. Frequency, window, standard errors, and residual-unit conventions were not supplied, so alpha and volatility are not annualized and negative residual signals are not converted into price forecasts. The unexplained portion includes noise and omitted influences; it cannot all be labeled company-specific information.

A more value-like return exposure does not establish undervaluation, and a small-size exposure does not make CoStar a small company. Similarly, relative stock performance against Zillow cannot prove customer migration. Operating evidence must establish competitive claims. The factor model helps frame risk sensitivity while the investment thesis remains a cash-flow argument.

Revalidated analytical lessons

Research-capital methods transfer as procedures, not as predetermined numerical outcomes. CoStar’s matched sensitivity raises its loss-stage ROIC, whereas several profitable software examples show lower returns after research capitalization. Neither result is universal. Guidance must likewise be compared on a constant acquisition perimeter, and compensation requires a consistent expense-or-dilution convention. [S3] [S4] [S5]

Adoption and attribution evidence establish relevance before incremental profit. Banks’ marked-capital rules and biotechnology acquired-research treatments do not automatically apply to this information portfolio. The analogous mortgage lesson belongs only in Zillow’s matched warehouse comparison. Verdict: the differentiated question is whether retained customer economics improve after restoring productive capacity, and whether the resulting cash remains available to shareholders rather than financing another indefinite investment cycle. [S2] [S17]

Fact vs. Interpretation

An accurately transcribed management statement is evidence of what management said, not independent confirmation of the economic claim. A reproducible analyst calculation can also remain assumption-sensitive. Verdict: the profitability inflection is better established than durable portal unit economics, acquisition returns, or realized AI savings.

Open Questions

The highest-value missing evidence concerns retained economics and capital returns. Questions should be prioritized by their ability to distinguish the competing explanations, rather than by the ease of producing another favorable activity measure. Current disclosures support the direction of cost improvement while leaving several product bridges incomplete. [S1] [S2]

1. Homes cohorts: What were gross additions, cancellations, first-anniversary renewals, and premium attach by cohort after the price changes? Flat membership cannot reveal replacement efficiency.
2. Price definitions: How do June subscriber price, September subscriber price, October new-sales price, recognized revenue, and run rate reconcile? Which customers receive promotional periods or different products?
3. Homes contribution: What are fully loaded acquisition cost, photography and support cost, premium contribution, and cash payback? What quarterly loss sits within Residential?
4. Apartments accounts: What are same-property spending, renewal, reactivation, and contribution by property size? How much inventory loss reflects completed leasing versus competitive displacement?
5. Selling capacity: How much shortfall arose from recruitment, attrition, deliberate reductions, or weak productivity? When do new field hires cover their complete cost?
6. Savings allocation: Which identified savings persist, which reverse with hiring, and which fund new products? What exact 2027 bridge supports the unchanged 2028 objective?
7. Acquisitions: How do Zonda consideration and net cash reconcile? What full-capital cash hurdles apply to Matterport, Domain, Zonda, and Vexcel?
8. Funding and obligations: What operating cash reserve will remain after Vexcel, what financing terms apply, and what construction or legal obligations remain?
9. Distribution and AI: What are direct, search, paid, and AI referral contribution economics? Can matched outcomes distinguish incremental leases and retained revenue from selection effects?
10. Governance: Will the board publish acquisition-cohort cash returns alongside organic revenue and adjusted EPS? What evidence would cause it to stop a project or further purchases?

The complete Q3 filing should improve purchase accounting, shares, liabilities, and legal visibility, but may still leave product cohorts undisclosed. Independent monitoring should therefore pair financial reconciliation with retained customer evidence. Verdict: a reconciled account, contribution, or acquisition-return bridge would change conviction more than a larger TAM or another engagement record. [S4] [S6] [S7]

What Must Be True

Bull-case requirements and falsifiers

The constructive premises begin with the verified commercial renewal and margin base, the $37 million bookings low, Residential’s first positive company-defined EBITDA quarter in the current sequence, and the disclosed rental-property decline. The thresholds below are analyst monitoring rules, not management guidance or statistically estimated failure probabilities. [S1] [S2]

Two consecutive positive Residential EBITDA quarters would improve durability evidence, but would still not prove Homes standalone economics. Organic growth above 8% with recovering bookings and paid inventory would be stronger than acquisition-led reported acceleration. The acquisition requirement was already challenged by Vexcel; its cash-return case now needs evidence rather than an assumed strategic premium. [S2] [S6]

Bear-case requirements and falsifiers

These bear falsifiers require more than a favorable housing cycle or one unusually strong quarter. A transaction recovery can temporarily lift several products while concealing weak acquisition payback. Conversely, an adverse macro quarter should not automatically invalidate a long-lived information franchise. Evidence should be compared on a constant perimeter and paired with the cost required to produce it. [S2] [S3] [S5]

The next checkpoints are the complete Q3 filing, Q4 bookings, a second Residential EBITDA observation, the 2027 budget, Vexcel’s expected Q1 closing and financing, and subsequent Homes premium renewals. Regulatory monitoring should verify court entry and implementation before dating Redfin’s rebuilding obligations. The sequence distinguishes information already available from future evidence needed to adjudicate the thesis. [S1] [S6] [S20] [S33]

Verdict: the operating thesis passed a profitability milestone while growth and capital returns remain unresolved. The decisive evidence must match bookings with selling cost, retention with monetization, adjusted profit with economic cash, and acquisition benefits with the full capital committed.

Evidence anchors: S1 — Q3 financial results, S3 — audited annual filing, S6 — Vexcel transaction disclosure, and S7 — governance and incentives.

Public source appendix

• S1: CoStar Q3 2026 results and financial tables — Primary issuer disclosure; unaudited financial tables; published 2026-10-27; Statements of operations, balance sheets and cash flows; quarterly and segment reconciliations; 2026 outlook.
• S2: CoStar Q3 2026 earnings call — Management commentary; transcript aggregation reconciled to issuer tables; published 2026-10-27; Florance and Rossmann prepared remarks; Q&A including Christiansen, Nagle, Thind and Huber. Speaker-labelled transcript retrieved through Company Financials.
• S3: CoStar 2025 Form 10-K — Primary audited SEC filing; publication date unavailable; Items 1, 1A, 5 and 7; financial statements pp. F-7–F-11; acquisition consideration pp. F-25–F-27; segment and EBITDA reconciliations pp. F-42–F-44; debt and commitments.
• S4: CoStar Q2 2026 Form 10-Q — Primary SEC filing; published 2026-07-29; Cover; accounting and commission policies; acquisition notes; debt; Brown and similar claims; MD&A subscription metrics; construction interest.
• S5: CoStar Q2 2026 earnings call — Management commentary; transcript aggregation; published 2026-07-28; Florance Homes pricing, run rate and cancellations; Lown guidance explicitly excluding Zonda. Company Financials speaker-labelled transcript.
• S6: CoStar proposed Vexcel acquisition disclosure — Primary issuer transaction announcement; published 2026-10-27; Price; trailing revenue and margin; cash/revolver funding; expected Q1 2027 close; large-customer concentration and AI execution risks.
• S7: CoStar 2026 definitive proxy — Primary governance and compensation filing; published 2026-04-30; Capital Allocation Committee termination; director retainers and equity; compensation redesign pp. 48–49, 56–57 and 61–64; 2025 incentive discussion.
• S8: CoStar Zonda closing announcement — Primary transaction disclosure; published 2026-08-21; Exhibit 99.1; $800m cash; $170m 2025 revenue and 23% adjusted margin; customers and products.
• S9: CoStar January 2026 capital-allocation and compensation filing — Primary SEC filing; published 2026-01-07; Items 5.02 and 7.01; tax gross-up removal, severance plan, repurchase and compensation announcements.
• S10: CoStar issuer stock information — Issuer-hosted market-data cross-check; publication date unavailable; Historical Stock Quote: October 27, 2026 close $32.01; issuer-hosted delayed data.
• S11: Company Financials — historical prices and financial diagnostics — Third-party market data and financial diagnostic; publication date unavailable; Two paginated CSGP responses, 1,254 observations; exchange-qualified Zillow and MSCI October 27 prices; profile, annual statements, profitability, enterprise value and historical multiples. Material inputs reconciled to filings.
• S12: CoStar 2024 Form 10-K — Primary audited SEC filing; published 2025-02-20; Consolidated statements, MD&A, development expense, campus purchases and equity.
• S13: CoStar 2023 Form 10-K — Primary audited SEC filing; published 2024-02-22; Consolidated statements; subscription/bookings definitions; development and marketing expense.
• S14: CoStar 2022 Form 10-K — Primary audited SEC filing; published 2023-02-22; Financial statements, revenue and expense comparisons, funded debt, equity and software development.
• S15: CoStar 2021 Form 10-K — Primary audited SEC filing; published 2022-02-23; Consolidated statements and MD&A; product economics and revenue comparisons.
• S16: Zillow Q2 2026 financial results — Primary competitor results; published 2026-08-05; Rentals growth 31% and multifamily growth 42%; financial and non-GAAP tables.
• S17: Zillow Q2 2026 Form 10-Q — Primary competitor filing; published 2026-08-05; Balance sheets; June ending-share reconciliation; mortgage loans and warehouse funding.
• S18: Zillow Q2 2026 shareholder letter — Primary competitor outlook; published 2026-08-05; 2026 revenue and adjusted-EBITDA outlook; compensation outlook; product and engagement commentary.
• S19: Google Home Listings Ads documentation — Primary competitor product documentation; publication date unavailable; Product operation, lead pricing, U.S. mobile availability, HouseCanary feed and Fair Housing restrictions.
• S20: Zillow–Redfin proposed stipulated order — Primary regulatory court submission; entry not established; published 2026-08-24; Document 400-1, proposed stipulated order; competition remedies and customer notice, including six months from court entry.
• S21: Rightmove H1 2026 results — Primary competitor results; published 2026-07-31; Financial highlights and operating-profit reconciliation: 69% underlying and 66% statutory operating margins.
• S22: Moody’s commercial real-estate capabilities — Primary competitor product description; publication date unavailable; Data, credit/risk and tailored CRE capabilities; historical coverage.
• S23: MSCI Real Assets Market Data — Primary competitor product description; publication date unavailable; Institutional real-estate transaction, benchmark, market and asset information.
• S24: Altus ARGUS products — Primary competitor product description; publication date unavailable; ARGUS Enterprise valuation and cash-flow forecasting; portfolio, development and fund products.
• S25: Andrew Florance August 2026 Form 4 — Primary insider transaction filing; published 2026-08-05; Table I: August 4 code-P purchase, 83,300 shares at $29.89; trading-plan checkbox.
• S26: CoStar CFO transition filing — Primary leadership filing; published 2026-07-13; Item 5.02; July 13 appointment effective July 31; Rossmann background and employment terms.
• S27: MSCI Q3 2026 Form 10-Q — Primary peer financial filing; published 2026-10-20; Cover: 71.5m October 13 shares; debt principal $6.760bn; cash including restricted; Q3 revenue $892m, adjusted EBITDA $554.8m and margins.
• S28: the factor model — October 27, 2026 — Internal statistical diagnostic; not fundamental evidence; published 2026-10-27; Supplied exposures, residual signals, R-squared and adjusted R-squared; frequency, window and uncertainty estimates absent.
• S29: Reuters report on Vexcel and after-hours trading — Reliable secondary market-event report; published 2026-10-27; October 27 article, 21:47 publication; $28.30 after-hours indication.
• S31: Andrew Florance May 2026 Form 4 — Primary insider transaction filing; published 2026-05-04; Table I: May 1 code-P purchases, 68,330 shares at $35.17 and 3,100 at $35.82.
• S32: Andrew Florance February 2026 Form 4 — Primary insider transaction filing; published 2026-03-02; Table I: February 27 code-P purchase, 55,720 shares at $44.52.
• S33: FTC announcement of proposed Zillow–Redfin order — Primary regulatory announcement; published 2026-08-24; Proposed-order description; court approval/signature condition; independent rentals rebuilding and continued syndication.
• S34: CoStar Q3 2026 investor presentation — Primary issuer presentation; published 2026-10-27; Slides 3–7: quarterly results, full-year and Q4 consolidated and segment outlook.