2026-08-06Internal (40-agent DeepSeek pass via OpenRouter + yfinance ground truth + 10-K/10-Q verification)
DEEP INTEL: NRDS, watching a company buy back the traffic it used to get free
Built from a 40-agent qualitative pass costing $0.029 in total, with every number then verified against the fiscal 2025 10-K, the first-quarter 2026 10-Q, SEC XBRL company facts and a yfinance harvest. The model contributed no figures by design and was told in every prompt that any number it emitted would be deleted. Its useful contributions were three: it flagged partner concentration as the thing to check, which turned out to be 26% of revenue in one customer; it flagged that revenue recognised on a variable, constrained estimate is not the same as revenue collected, which the 10-K confirms; and it asked whether marketing costs were being capitalised, which sent me to the note confirming advertising of $486.4M is expensed as incurred. Setup: $122.8M of trailing free cash flow against a $575.7M market capitalization, no corporate debt, 93.0% gross margins, and a marketing line that shows the company substituting paid traffic for the organic traffic AI is taking. Conviction 3/5.
2026-02-25NerdWallet FY2025 Form 10-K, sales and marketing note
The three-line table that explains the whole stock
The 10-K breaks sales and marketing into three components and the trend in them is the investment case, in either direction. Performance marketing, which is primarily the cost of driving traffic directly to the platform, was $206.5M in 2023, $297.4M in 2024 and $416.9M in 2025. Brand marketing went the other way, $84.4M, $68.6M, $65.1M. Organic and other marketing, which is primarily personnel cost for the content and marketing teams, went $110.6M, $104.6M, $102.7M. Total sales and marketing was $401.5M, $470.6M and $584.7M. Advertising expense, which includes performance and brand marketing plus certain external organic costs, totalled $294.7M, $370.3M and $486.4M and is expensed as incurred rather than capitalised. Set the performance line against revenue of $599.4M, $687.6M and $836.6M and the ratio moves from 34.5% to 43.3% to 49.8%. Revenue grew 39.6% over those two years; the cost of buying the traffic behind it grew 102%.
2026-02-25NerdWallet Q4 and FY2025 results, chief executive statement
Management says the quiet part on the record
Chief executive Tim Chen, in the fiscal 2025 release: 'Strength in performance marketing, direct, and non-search referral channels more than offset organic search headwinds as consumers turned to AI overviews and LLMs over traditional search.' The same release reports fourth-quarter credit cards revenue of $26.5M, down 24% year over year, and SMB products revenue of $22.5M, down 12%, both attributed to continued headwinds in organic search traffic, against loans revenue up 141% to $42.3M and emerging verticals up 57% to $52.9M as partners expanded budgets. Full-year revenue was $836.6M, up 22%, with GAAP operating income of $65.2M against $9.4M. The candour is worth something. A company that names the mechanism disassembling its own funnel, in its own results release, is easier to underwrite than one that attributes the same decline to a temporary algorithm update.
2026-05-06NerdWallet Q1 2026 results and Form 10-Q
The first guidance miss, and what grew anyway
First-quarter 2026 revenue was $222.2M, up 6% year over year, against management's own guidance of $224-232M issued ten weeks earlier. Underneath it: consumer revenue of $197.6M rose 10%, driven by a $20.9M increase from deposit accounts and $12.7M from personal loans as partners expanded budgets, partially offset by a $12.5M decrease in consumer credit cards attributed to 'continued pressures in organic search traffic that have persisted for multiple quarters'. SMB revenue of $24.6M fell 15% for the same stated reason. GAAP income from operations was $27.2M against $0.7M, net income $20.4M against $0.2M, and adjusted EBITDA $45.2M against $21.1M. The reason profit exploded while revenue crawled is on the expense side: sales and marketing FELL 7% to $149.1M while revenue rose 6%, the first quarter in this sequence where the marketing line did not outrun the top line. Cash fell from $98.3M to $56.3M, but that is the $66.0M share repurchase and the $16.9M College Finance acquisition, not operating burn.
2026-05-06NerdWallet Q1 2026 Form 10-Q, notes 3, 4, 5 and 7
Two things on this balance sheet that will fool a screener
The first is the mortgage warehouse. NerdWallet's lending subsidiary maintains a $15.0M warehouse line of credit, extendable to $18.75M for up to 90 days, bearing interest at the rate of the underlying mortgages subject to a 5.25% floor and maturing February 1 2027. It had $14.9M outstanding at March 31 2026 against $6.9M at December 31 2025, recorded inside accrued expenses rather than as debt, and it funds $15.2M of mortgage loans held for sale carried at fair value. The two offset almost exactly. Neither is corporate leverage, and the company's own adjusted free-cash-flow definition adds back the net warehouse borrowing for precisely this reason. The corporate revolver, a $125.0M senior secured facility with JPMorgan maturing September 2028, had nothing drawn at either fiscal year end. The second trap is the share count. There are 34,113,131 Class A shares and 31,685,652 Class B, so 65,798,783 in total, and several data feeds publish only the Class A number. At $8.75 the correct market capitalization is $575.7M; using the Class A count alone would produce $298.5M and make every valuation ratio on this page look roughly twice as attractive as it is. Also in the quarter: the February 20 2026 acquisition of College Finance Company, a student loan marketplace, for $17.2M of consideration of which $16.9M was cash.
2026-06-30NerdWallet Form 4 filings, SEC EDGAR
The best-informed seller in the record
Topline Capital Management and Topline Capital Partners, reporting as a ten percent owner, sold on three consecutive trading days at the end of June 2026: 919,153 shares at $8.95 on June 26, 587,835 at $9.22 on June 29, and 454,872 at $9.27 on June 30. Shares held following each transaction were 4,331,992, then 3,744,157, then 3,289,285, against 5,251,145 before the sequence began. That is 1,961,860 shares for approximately $17.9M and a 37.4% reduction in the position, executed roughly $1.50 above the fifty-two-week low of $7.33 and about 43% below the fifty-two-week high. The parsed Form 4 record contains no open-market insider purchases over the same window. Set against that, the company itself repurchased $66.0M of stock during the first quarter of 2026, its largest quarterly buyback on record. The issuer and its largest reporting outside holder are taking opposite sides of the same disagreement, which is a reasonable summary of why the multiple is where it is.
2026-08-06yfinance and SEC XBRL (computed, USD)
What the market is paying, and one number this page refuses to print
At the August 6 2026 close of $8.75 the market capitalization on 65,798,783 shares is $575.7M. Cash was $56.3M at March 31 2026 with nothing drawn on the corporate revolver, so enterprise value is about $534.4M taking the $14.9M warehouse borrowing at face value, or about $519M if the matching mortgage loans held for sale are netted against it. Trailing twelve-month revenue is $849.6M, net income $68.9M, operating income $91.7M and free cash flow $122.8M, so the market is paying roughly 4.4 times trailing free cash flow for the operating business and the free-cash-flow yield on the market capitalization is 21.3%. Trailing price to earnings is 9.4 on trailing diluted earnings of $0.93. The shares are 46.1% below the $16.24 fifty-two-week high, with the 200-day moving average at $11.03 and short interest at 13.4% of the Class A float. Five analysts publish price targets with a mean and median of $12.00, a low of $9.00 and a high of $15.00. This page deliberately does not publish a forward price-to-earnings ratio. The widely-syndicated figure of about 4.3 times is computed against a fiscal 2027 earnings estimate carrying a single contributing analyst, and that estimate appears to be a non-GAAP number being divided into a GAAP share price. Management's own fiscal 2026 guidance of $72-89M in GAAP operating income implies GAAP earnings per share somewhere near $0.82 to $1.01, which is a forward multiple closer to 9 or 10 times than to 4. Publishing the low number would have made this page look twice as cheap as the company's own guidance supports.