Here’s a pairing that should bother anyone who owns quality compounders. CSGP runs a commercial real estate data franchise (CoStar Suite, LoopNet) with a roughly 34% EBITDA margin, quarterly renewal rates that have climbed from 89% to 92–94%, and a genuine cyclical tailwind behind it: CBRE just reported CRE investment volume up 19% YoY, with its lending index at the highest level since 2021. That’s about as good as subscription data businesses get. And the stock is at $28.39, down about 58% this year, roughly 70% off its 52-week high of $97, and it kept falling 21% after an April quarter that beat guidance and came with a raise.The macro didn’t do this nor Short sellers as short interest is under 4% of float. What did this is what management chose to do with the cash. CoStar took a moat-protected franchise the market used to pay 8.5x to 12.85x trailing revenue for and spent years funding a land grab against Zillow with it. The result is a stock that has destroyed enormous shareholder value on the way down and, frustratingly, still isn’t a screaming buy at the bottom. That second part is worth walking through carefully, because the bull case at these levels sounds pretty good until you press on it.The bull caseAt $28.39, the longs have real arguments. The stock trades around 3x forward revenue against that 8.5x–12.85x historical range (trailing basis, so not a perfect comparison, but the gap is enormous). Residential segment losses are narrowing fast, from -$85M a year ago to -$29M in Q1, with management guiding to breakeven-or-better in Q2. Sell-side EPS estimates have actually drifted up 1–5% over the past 90 days while the price fell 21% — the numbers aren’t capitulating, just the multiple. The mean price target sits at $47.50. There’s a $700M buyback running, $505M of it already executed in Q1, which was the first quarter the share count actually declined. And the Commercial segment’s recovery is corroborated by CBRE’s independent data, not just management’s framing. Each of these is factually true but they add up to less than they appear.Cheap on the wrong metricThe 3x-revenue argument compares today’s CoStar to a company that no longer exists. The Matterport and Domain acquisitions permanently diluted the margin structure, and SBC ran at 6% of revenue last year. On the lens this business should actually be valued on EV/adjusted EBITDA and CSGP trades at 14.2x. In other words, the stock is trading at its own terminal assumption. That’s not cheap and is a fair multiple with no margin of safety attached.The sum-of-the-parts math is more damning and indicative as to what the market thinks of the residential business. Commercial should earn roughly $684M of EBITDA this year. Put a 16x multiple on that, which is defensible for a business with 92–94% renewal rates, and Commercial alone accounts for about $10.9B of the $11.4B enterprise value. Everything else, meaning Apartments.com, Homes.com, Domain, and the roughly $3B of goodwill allocated to Residential, is being valued at around $400M. Use 18x, the kind of multiple Verisk or MSCI-class data businesses command, and Residential’s implied value goes negative.So the bull has a problem, the more premium you argue the crown jewel deserves, the more clearly the market is telling you it thinks the Residential bet is worthless. That’s not a market failing to notice a free option but a market that looked at the option and priced it at zero.The goalposts keep movingThe Q2 breakeven guide is the hinge of the near-term bull case, and it comes from a management team with a documented pattern of redefining success right before missing it. The 50% unaided-awareness target was missed at 42% and then dropped from disclosure entirely. NPS for the same quarter was reported as 43 on one call and restated as 9 the next, without explanation. Homes.com’s full-year 2025 visitor count was cited as 100M in February and 108M in April, for the identical period. Five-year targets set in 2022 were formally withdrawn in 2025. Each substitution arrived just as the prior metric was about to embarrass.And the current milestone has a definitional sleight to it. “Residential segment breakeven” is an easier claim than it sounds, because the segment includes Apartments.com, which is already profitable and subsidizes Homes.com inside the blend. Management’s own January framework, never walked back, says Homes.com specifically won’t be profitable until 2030. Clearing the Q2 bar would prove the blended segment can stop bleeding. It proves nothing about whether the actual bet works.What the money actually boughtThis is the part that generalizes beyond CoStar. Strip out the adjusted-EBITDA framing and look at cash. SBC-adjusted free cash flow was negative $334M in FY24 and negative $153M in FY25. The FY26 projection turns positive by all of $16M, a 0.4% margin, which is noise. Homes.com alone absorbed roughly $850M of net investment in 2025 (you can back that out of management’s own guide math), with $550M more planned this year.The buyback record is worse than it looks, too. CoStar spent $575M on repurchases in FY25 and the share count went up 2.1%, because $881M of Matterport stock issuance swamped it. Shareholders were paying to buy back stock and getting diluted anyway. Q1 2026 was the first quarter the mechanics actually worked, which makes it one data point, from a company that has done two stock-funded deals in 24 months.Add it up: billions committed to Residential in cash burn, stock issuance, and goodwill, and the market’s current appraisal of what all that bought rounds to zero. Meanwhile even the crown jewel has developed a crack as CoStar Core net new bookings have declined three straight quarters, from $93M to $67M, over the exact window CBRE says the CRE macro was accelerating. When an analyst raised the sequential trend directly on the Q1 call, management answered with the year-over-year number instead. That’s the leading indicator for the one segment everyone agrees is great, diverging from its own end market, and it’s the question management least wants to engage.So what do you do at $28.39?Separate the business from the stock. The business is a genuinely excellent Commercial franchise attached to a capital-allocation program that has consumed years of its output. The stock, at 14.2x EBITDA with a 0.1% real FCF yield, needs a parlay to work from here: Residential clears breakeven in Q2, bookings reaccelerate in the seasonally strongest quarter, SBC and capex discipline both hold, the buyback keeps outrunning dilution, and the AI-search fear that has taken Zillow down 53% on the identical theme proves overstated. Any one of those failing sends this toward the house model’s perpetuity floor of $19, and the bear case pencils to the low-$20s or worse. Note also that CSGP still trades at a premium forward P/E to Zillow (about 16x vs. 10.5x) even though Zillow is already profitable at the consolidated level.There is a real upside scenario. The market is pricing an implied terminal multiple of about 9x against the current 14.4x, and a clean July 28 print with Residential breakeven and a bookings reacceleration, with management keeping Residential and Homes.com language honestly separate and clean could close a meaningful piece of that gap. If both bars clear, this gets interesting. But that’s a trade on an event, not a compounding thesis, and it requires trusting a scoreboard that has been rewritten three times.The larger lesson is the one worth keeping. CoStar shows how little a great core business protects you when the cash it generates gets deployed badly. The moat was real the whole way down. Shareholders still lost 58% in six months, got diluted through their own buyback, and now hold a stock where the market values a multi-billion-dollar strategic bet at nothing. Business quality isn’t the thesis. Capital allocation is.July 28 will tells us a lot. I’d want to see both metrics clear before paying up for this, and even then, the ROIC Homes.com question doesn’t actually get answered until 2029+.