Financial Shenanigans

Financial Shenanigans — The Trade Desk, Inc. (TTD)

Forensic verdict: Watch (score 38/100), sitting at the top of the band. The Trade Desk's books are among the cleaner ones in ad-tech: revenue is recognized conservatively on a net (agent) basis, cash flow is genuine cash rather than a payables trick, there is no acquisition roll-up to disguise, and Ernst & Young — auditor since 2015 — has issued clean opinions with a single, well-telegraphed Critical Audit Matter [1]. The risk here is not fabricated earnings — it is optics and governance: a business whose headline cash generation is roughly two-thirds funded by adding back ~$490 million of annual stock-based compensation, a balance sheet that dwarfs the income statement because TTD passes ~$13 billion of client spend through it, a one-time deferred-tax tailwind flattering FY2025 cash flow, and a founder-controlled governance structure carrying an $819 million disputed CEO option, a Nevada re-domicile, a revolving cast of CFOs, and live securities litigation.

The two things to watch are (1) the stock-based-compensation-funded cash-flow narrative and (2) the breeding ground — founder super-voting control, the mega-option, and finance-leadership turnover. The cleanest offsetting fact: operating cash flow is not manufactured from working capital — receivables actually grow slower than revenue and TTD funds a net receivable float to its own clients, so cash conversion is real. The single data point that would most move the grade: whether, under a new finance team, the revenue-recognition Critical Audit Matter widens the "gross-basis Supplier Components" scope or the non-GAAP/DSO definitions drift.

Sources: score and flag counts are this analysis's assessment; CFO/NI, FCF/NI, accrual ratio and the Adjusted-EBITDA-to-net-income gap are derived from reported financials, FY2023–FY2025 Statements of Cash Flows and Operations [2]; Adjusted EBITDA reconciliation [3].

The intermediary lens — why the balance sheet looks nothing like the P&L

Read this first: everything else follows from it. TTD is a demand-side platform that sits between advertising agencies and media suppliers. In FY2025 roughly $13.4 billion of client spend ("gross spend") flowed across the platform, but only $2.90 billion was recognised as revenue — TTD keeps a platform fee of roughly one dollar in five and passes the rest through [4]. Because TTD invoices agencies for the gross amount and owes suppliers the gross amount, its receivables and payables are sized to the ~$13 billion flow, not the ~$2.9 billion of revenue.

Source: FY2025 10-K, MD&A gross-spend table [5] and Consolidated Balance Sheets [6].

The forensic consequence: standard balance-sheet screens misfire on TTD. Accounts receivable of $3.77 billion is 130% of net revenue, which would flag as a five-hundred-day DSO on any naive screen — but that is an artifact of gross billing, not slow collection or channel-stuffing [7]. Any DSO, accrual, or receivables-to-revenue metric on TTD must be read against gross spend, not revenue. This same structure is the reason the SEC challenged TTD's gross-vs-net presentation before the IPO (see Breeding ground), and it is why revenue recognition remains the auditor's sole Critical Audit Matter today.

Cash-flow quality — real cash, but SBC-funded and partly one-time

Takeaway: the cash is genuine, but the "quality" flattering it is stock-based compensation add-backs and a one-off deferred-tax swing, not operating leverage — and it is not a payables trick. Over FY2023–FY2025 operating cash flow ran 2.3x reported net income and free cash flow 2.0x net income — the kind of spread that should trigger a "why is CFO so far above earnings?" investigation.

Source: FY2025 10-K Consolidated Statements of Cash Flows (FY2023–FY2025) [8]; earlier years company filings, as reported.

The mechanism is unambiguous once you decompose the FY2025 bridge from $443 million of net income to $993 million of operating cash flow. The single biggest reconciling item is $490.6 million of stock-based compensation — larger than net income itself — followed by a $167.7 million deferred-income-tax add-back and $115.8 million of depreciation. Working capital was a net drag of roughly $275 million, not a source [9].

Source: FY2025 10-K Consolidated Statements of Cash Flows [10].

Three findings fall out of this bridge:

1. The deferred-tax add-back is a non-recurring tailwind (yellow). Deferred income taxes swung from reducing cash flow by $76.9 million in FY2024 and $61.6 million in FY2023 to adding $167.7 million in FY2025 — a ~$245 million year-over-year swing that alone explains most of FY2025's 34% CFO growth. It is driven by the July 2025 OBBBA tax law, which let TTD recognise $175 million of domestic R&D costs as an income-tax receivable with a matching reduction in deferred tax assets (the balance-sheet deferred-tax asset fell from $230 million to $56 million) [11]. This is legitimate, but it is a one-time cash-flow lift that will not repeat, and it makes FY2025 CFO growth look better than the underlying run-rate.

2. Working capital is a drag, not a lifeline (green — clean). The classic "CFO propped up by stretching payables" shenanigan is absent here. In FY2025 the AR build (−$433 million) exceeded the AP build (+$291 million), a net use of cash — because, by TTD's own risk disclosure, it "generally pay[s] our accounts payable on shorter cycles than we collect on our accounts receivables, requiring us to remit payments from our own funds" [12]. TTD funds float to its clients rather than living off supplier float. Receivables have grown in line with or slower than revenue every year, further evidence there is no collection-timing game.

Source: FY2025 10-K Consolidated Balance Sheets and prior-year filings [13]; earlier years company filings, as reported.

3. There is no financing-into-operating or acquisition-into-operating distortion (green — clean). TTD carried no drawn debt at year-end, borrowings under its revolving facility are shown in financing, its lone FY2025 acquisition (Sincera) cost $4.4 million of cash and sits in investing, and a prior-year cash-flow reclassification was limited to aggregating the credit-loss provision line with "no impact to cash flows from operating, investing or financing activities" [14]. CF1, CF2, and CF3 all test clean.

Earnings quality — conservative recognition, rate-sensitive income, thin reserves

Takeaway: recognition is conservative and there is no capitalization game, but two yellow items deserve tracking — the rate-sensitivity of "other income" and a very thin credit-loss allowance.

Revenue recognition is conservative (green, with a caveat). TTD recognises revenue at a point in time, when a bid is won and the client's purchase clears the platform, and books the platform fee net (as agent) — it only grosses up the narrow category of "Supplier Components" (value-added services and data where it controls the service) [15]. Net presentation is the conservative choice; the risk is one-directional — that the gross-basis Supplier-Components scope quietly widens — which is exactly why E&Y flags revenue recognition as its Critical Audit Matter, citing "the high degree of audit effort in performing procedures related to client purchases" [16]. No pull-forward is visible: receivables grew 13% against 18% revenue growth in FY2025.

"Other income" is rate-sensitive, not operating (yellow). With a securities-and-cash portfolio around $1.3 billion, interest income has been a meaningful and rate-cycle-dependent contributor to pre-tax income — peaking near 34% of operating income in FY2023 before fading as rates and the cash balance fell. Adjusted EBITDA correctly deducts the $68.7 million of net interest income, but GAAP pre-tax income and EPS carry it, so a chunk of "profit growth" in 2022–2023 was the Fed, not the platform [17].

Source: derived from reported financials, FY2021–FY2025 Statements of Operations [18].

No expense-shifting via capitalization (green). TTD capitalises almost nothing: capitalised software development costs were just $12.8 million in FY2025 against ~$490 million of expensed technology spend, so there is no "route R&D into an asset" game. Capital expenditure did surge — $47M → $98M → $197M across FY2023–FY2025 — but that is a genuine data-center build-out, disclosed as such, and depreciation is rising behind it (D&A up from $80M to $116M). Capex running ahead of depreciation is expected during a build phase; it is worth tracking, not a red flag, but note that $104.5 million of FY2025 capital assets were financed through accounts payable (a non-cash item), so the cash capex line understates the commitment [19].

Source: FY2025 10-K Consolidated Statements of Cash Flows [20]; earlier years company filings, as reported.

Under-reserving is a low-grade yellow. The allowance for credit losses is $12.2 million against $3.77 billion of gross receivables — 0.32% [21]. That looks thin, and it is partly defensible: "sequential liability" terms mean many agency receivables are not TTD's credit risk until the advertiser pays. But the allowance has barely moved ($11.2M → $12.2M) while receivables have more than doubled since FY2021, and the CTV/agency concentration is rising (below). A reserve that never grows into a fast-growing, concentrated book is a place a future problem would hide.

No big bath (green). The December-2024 reorganization that accompanied the Q4-2024 revenue miss was framed as an operational reset "to accelerate opportunities across CTV, retail media, identity, supply chain optimization, and audio" [22]. Critically, it produced no discrete restructuring charge and no asset write-off — severance was expensed within operating expense, and TTD states "no material impairments have been recorded" on its long-lived and intangible assets [23]. There is no evidence of kitchen-sinking a weak quarter to lower a future base.

Metric hygiene — stock-based comp is the whole game

Takeaway: the single largest forensic issue on the page is metric optics — adjusted profitability and "free cash flow" are both heavily subsidised by adding back real, recurring, dilutive stock compensation. This is the strongest yellow flag.

In FY2025, GAAP net income was $443 million but Adjusted EBITDA was $1,196 million — a $753 million gap, of which stock-based compensation ($491 million) is roughly 65% [24]. SBC is not an unusual or "one-time" cost for TTD — it has run remarkably flat at ~$490 million for three straight years and is the primary tool the company uses to pay its people. Adding it back to reach an "operating" metric, then also adding it back inside operating cash flow, means the same ~$490 million is excluded twice from the numbers investors anchor on.

Source: FY2025 10-K Statements of Cash Flows (SBC and FCF) [25].

Stated as a ratio, SBC equalled 89% of free cash flow in FY2023 and still 62% in FY2025. The trend is genuinely improving as revenue scales (SBC fell from 25.3% to 16.9% of revenue), which is why this is a yellow and not a red. But a PM should treat TTD's ~$796 million of FY2025 free cash flow as roughly $305 million once the economic cost of the equity used to generate it is respected — and note that TTD spent $1.38 billion on buybacks in FY2025, more than its entire free cash flow, largely to mop up that dilution and shrink the share count; that buyback drained cash from $1.37 billion to $658 million and pushed the company to a small accumulated deficit [26].

Source: FY2025 10-K Adjusted EBITDA reconciliation [27].

"Gross spend" is a scale-flattering metric, but honestly labelled (yellow). Management leads with $13.4 billion of gross spend — a number 4.6x revenue — "to assess our market share and scale" [28]. It is disclosed with a clear reconciliation and caveats, so it is not deceptive — but it is the same gross-flow framing the SEC pushed back on pre-IPO, and it lets TTD present a far larger scale figure than its actual take. Balance-sheet metric distortion (KM2) is structural, not managed: the concentration disclosure shows two agency holding companies accounted for 30% of gross billings in FY2025 (each above 10%), up from one holding company above 10% in FY2024 [29]. Rising customer concentration against a static, thin allowance is the intersection to monitor.

Breeding ground — founder control, an $819M option, a re-domicile, and a CFO carousel

Takeaway: the governance breeding ground is the elevated part of this file and it amplifies, rather than dampens, the metric-optics risk — because the people who define the adjusted metrics also control the votes.

Founder super-voting control. Class B shares carry ten votes to Class A's one; as of December 31, 2025, Class B holders — including executives, directors and their affiliates, led by founder-CEO Jeff Green — held ~49.9% of total voting power, enough to influence or control most shareholder matters, with the dual-class structure not sunsetting until December 2035 [30]. Concentrated founder control with limited independent challenge is a classic breeding-ground condition.

The $819 million CEO Performance Option. In October 2021 the board granted Green a market-based option over up to 16 million shares (up to 19.2 million with a TSR modifier), exercise price $68.29, price hurdles from $90 to $340, grant-date fair value ~$819 million, expensed graded-vesting over ~five years [31]. This one grant drove $198M, $128M and $67M of SBC in FY2023–FY2025 respectively — i.e. it is a large slice of the very SBC add-back that inflates Adjusted EBITDA and CFO, tying the governance flag directly to the metric-hygiene flag. The option has been the subject of shareholder litigation, and TTD's November 2024 reincorporation from Delaware to Nevada — an anti-takeover-tightening move — is widely read as related to that dispute.

Live litigation. Following the Q4-2024 miss and stock drop, securities class actions were filed in February and March 2025, alongside consolidated data-privacy/wiretapping suits — legal exposure that management notes will make G&A "fluctuate period to period" [32]. These are disclosed as reasonably-possible with no material accrual, consistent with early-stage cases, but they are a governance-and-contingency watch item.

Finance-leadership turnover (yellow). The CFO seat has changed hands repeatedly across the corpus — Blake Grayson, then Laura Schenkein, then Alex Kayyal, then Tahnil Davis — and FY2025 G&A includes "an acceleration of stock-based compensation in connection with an executive transition" [33]. Repeated turnover in the office that owns the non-GAAP definitions and the DSO/DPO narrative raises the odds of a definitional drift and is the reason "watch the metrics under the new team" is the grade-moving variable.

The one already-settled negative — historical, and remediated. Before the IPO, the SEC's 2016 comment letters directly challenged TTD's gross-vs-net revenue presentation, its gross-spend metric, agency-customer concentration, and a restatement of 2014 financials tied to identified material weaknesses in internal control [34]. This matters as context — the same pressure points (gross vs net, concentration) persist structurally today — but it was resolved a decade ago, the company IPO'd, and there has been no restatement, auditor change, or material-weakness disclosure in the public-company era since. That clean subsequent record is a major reason the score stays in "Watch" rather than "Elevated."

The 13-category scorecard

Every shenanigan category, graded. Only three carry a live yellow; the rest test clean or benign. The concentration is in metric hygiene and one-time items, not in the core statements.

Sources: severity/evidence derived from FY2021–FY2025 10-Ks — Statements of Cash Flows [35], Balance Sheets [36], Adjusted EBITDA reconciliation [37], revenue-recognition note [38] and concentration note [39].

What to underwrite next

Five specific, named diligence items — not generic caveats:

1. Discount free cash flow for SBC. Treat reported FCF (~$796M FY2025) net of stock-based compensation (~$491M) — a "cash FCF" nearer $305M — when setting a valuation multiple or FCF yield. Track SBC/revenue: continued decline (from 16.9%) confirms the improving trend; a reversal is a downgrade trigger.
2. Normalize FY2025 CFO growth for the OBBBA deferred-tax lift. The $168M deferred-tax add-back and $175M R&D tax receivable are one-time; recompute the FY2026 CFO/NI ratio without them before extrapolating 34% CFO growth.
3. Watch the revenue-recognition CAM and the "Supplier Components" gross-basis scope. A material widening of what is recognised gross (vs the current net platform fee) would inflate reported revenue growth without more economic activity — read Note 2 in each 10-K/10-Q for scope language.
4. Watch the allowance-to-receivables ratio against rising concentration. $12.2M (0.32%) against a book where two holding companies are now 30% of gross billings. A provision that stays flat while concentration climbs, or a single large advertiser default, is where a loss would surface.
5. Watch non-GAAP and DSO/DPO definitions under the new finance team. After repeated CFO turnover, any change in Adjusted EBITDA add-backs, the gross-spend definition, or the DSO/DPO framing in the MD&A/press releases is a red-flag escalation.

Signal that would DOWNGRADE the grade (toward Elevated/High): the gross-basis revenue scope widens materially; a non-GAAP/gross-spend/DSO definition changes without clear reconciliation; the credit-loss allowance proves inadequate on a concentrated default; or securities litigation moves to a material accrual.

Signal that would UPGRADE the grade (toward Clean): SBC/revenue keeps falling toward peer norms; the CEO Performance Option finishes vesting (fully recognised by Q1 2026) and SBC steps down durably; finance leadership stabilises; and interest-rate-independent operating income keeps carrying the profit growth.

Decisive read: the accounting risk at The Trade Desk is a valuation haircut and a metric-discipline caveat, not a thesis breaker. The statements are conservative, the cash is real, and there is no restatement, auditor, or capitalization red flag. What a PM must not do is take Adjusted EBITDA or reported free cash flow at face value — both are meaningfully subsidised by ~$490M of annual, recurring, dilutive stock compensation that management then spends more than all of FCF buying back — and must weigh a founder-controlled governance structure that concentrates power over exactly the metrics and disclosures where the discretion lives. Size the position for optics and governance risk; do not underwrite it for fraud risk.