Financial Shenanigans

Financial Shenanigans — Yatsen Holding Ltd (YSG)

Figures converted from CNY at historical FX rates — see data/company.json.fx_rates. Ratios, margins, and multiples are unitless and unchanged.

The Forensic Verdict

Yatsen is a Watch with two concentrated yellow flags and a generally clean cash-flow signature for a still-loss-making company. The forensic risk score is 38/100. The two issues that matter are (1) cumulative goodwill impairments on the FY2020–early-FY2021 skincare acquisition stack — US$49.9M in FY2023 and US$55.2M in FY2024, concentrated in the Eve Lom reporting unit — which were disclosed correctly but reveal the price paid for the strategic pivot was too high ([1]; [2]; [3]); and (2) growing related-party inventory and service purchases from companies over which Yatsen "exercises significant influence" [4]. Offsetting these: the FY2022 SDNY securities class action filed in 2022 was dismissed with prejudice in 2025 and the case closed [5], DSO has collapsed from 88 days (FY2018) to 19 days (FY2025) with receivables actually shrinking on a 27% revenue jump, internal controls over financial reporting were assessed effective as of year-end [6], and FY2025 CFO landed essentially at GAAP net loss — i.e. zero accrual gap. The data point that would most move the grade is whether the next two years of related-party purchase disclosure show the same scale and counterparties, or whether the dollars migrate to Mr. Huang's directly-controlled vehicles (which would push this into Elevated).

Forensic Risk Score (Watch)

Red Flags

Yellow Flags

3y CFO / Net Income

0.29

3y FCF / Net Income

0.38

Accrual Ratio FY2025

0.004

Recv Growth − Rev Growth (pp, FY25)

-27.9

Goodwill+Intangibles / Total Assets

18.0%

No Results

Net forensic read: the accounting tells a credible story of a heavily over-marketed brand portfolio that was over-acquired in 2020-21, wrote down the difference in 2023-24, and is now running on cleaner economics. There is no evidence of revenue manufacturing, channel stuffing, or cash-flow dressing. The legitimate concerns are governance concentration, related-party purchase scale, and inventory days creeping up.

Breeding Ground

The breeding-ground profile is founder-dominated but professionally audited, which is the most common configuration for a Chinese ADR after a wave of accounting scandals. The dual-class structure giving Class B shares 20 votes versus 1 for Class A is set out in the company's Memorandum of Association ([7]; [8]). Two structural items raise the baseline risk; three offsets dampen it.

No Results

The breeding ground does not amplify the accounting findings the way a typical fraud setup would: there is no streak of meeting-beating, no aggressive guidance culture (management has guided cautiously since FY2022), no auditor change, no late filing, and no internal-control deficiency — management opined ICFR effective at FY2022 and again at FY2024 ([9]; [10]). Bonnie Yi Zhang was named audit committee chairwoman and Alan Hao Zong was appointed independent director per the Mar-26 board change disclosure [11]. The Nov 2023 NYSE non-compliance letter (ADS below US$1 average for 30 consecutive trading days) is on file [12]; an earlier round in 2022 was cured by August [13]. The dual-class voting structure means independent directors cannot vote Mr. Huang out, but they can — and have — kept the audit and reporting machinery in good order. The reader's appropriate posture: assume capital allocation decisions (acquisitions, buybacks, related-party arrangements, the new convertible notes deal) reflect the founder's view, not the consensus of a balanced board.

Earnings Quality

Earnings quality is better than the loss line suggests because the operating loss has been carrying a heavy load of non-cash and non-recurring charges — goodwill impairments and intangibles amortisation primarily — that are properly disclosed and now tapering. FY2025 had no goodwill impairment, and the gross-margin path (63% → 78%) is corroborated by the documented mix shift from Color Cosmetics (lower margin) to Skincare Brands (higher margin), not by pricing aggression.

The goodwill stack tells a coherent — and unflattering — story. Yatsen acquired Galénic, the mainland China business of DR.WU and Eve Lom in late 2020 / early 2021 [14], taking goodwill from US$3M to US$137M and intangibles from US$29M to US$117M in a single year. By FY2023 management began testing the Eve Lom reporting unit and recorded RMB354.0 million (US$49.9 million) of goodwill impairment [15], then a further RMB403.1 million (US$55.2 million) in FY2024 [16]. The FY2025 goodwill rollforward confirms zero further impairment and a remaining goodwill balance against the DR.WU reporting unit [17]. This is a trickle big-bath — split across two years against shrinking revenue — rather than a single shock, which is the harder pattern to time but the more honest accounting outcome. There is no evidence that operating margins were polished by stripping costs into the impairments; the SG&A ratio, fulfilment ratio, and gross margin all moved monotonically and explicably.

The non-cash adjustment burden was US$58M in FY2023 and US$71M in FY2024 — about 12% and 15% of revenue, respectively — and collapses to US$9M (1.4%) in FY2025 as the goodwill is largely gone and intangibles amortisation steps down (the Eve Lom intangibles are now nearly fully amortised; only US$6.1M was charged into selling and marketing in FY2025 vs. US$14.6M in FY2024 per the MD&A). This is the engine of the reported FY2025 turnaround: the headline operating loss contracted from −US$113M to −US$27M, and roughly US$65M of that US$87M improvement is the absence of the prior-year goodwill charge plus the ~US$9M drop in intangibles amortisation. The skincare brand growth (+63.5% YoY) is real and segment-level losses did narrow, but the FY2025 inflection is majority a non-cash-charge tailwind, which has to be flagged for investors using the slope of earnings to extrapolate.

Receivables and revenue tell the cleanest story on this page.

DSO improved from 88 days in FY2018 (an outlier because of the small revenue base) to a tight 19 days in FY2025 — consistent with a DTC channel mix that recognises revenue at delivery and collects through the e-commerce platform. There is no evidence of selling to channel to dress quarter-end. DIO, by contrast, has been steadily worse since FY2020 and reached 198 days in FY2025 — high for a fast-moving beauty business and inconsistent with the marketing message that Skincare is selling through. Inventory grew US$18M (+32%) in FY2025 against a +27% revenue lift, so it is not unambiguously stale yet, but management's own MD&A flags an "increase of US$17.9M in inventories" as the largest single working-capital drag on FY2025 CFO. Watch this in the FY2026 numbers — if DIO crosses 220 days without a coordinated promotional reset, an inventory write-down becomes a real possibility.

Cash Flow Quality

Operating cash flow has been negative every year since IPO except FY2022 and FY2025 conversion was essentially identical to net income — −US$13.5M CFO vs. −US$13.2M net loss, an accrual ratio of 0.4 bps of average assets. There is no evidence of CFO inflation; if anything, the company is owning the cash loss rather than managing toward it.

Two cash-flow vignettes deserve attention. First, FY2019 (pre-IPO) showed US$11M of net income but −US$1M of CFO — a classic earnings-without-cash signal that was central to the now-dismissed Maeshiro complaint. The court was unpersuaded the gap rose to securities fraud; it does, however, document that the IPO was priced on earnings that did not generate cash. Second, FY2022 was the only positive-CFO year (+US$20M) — but revenue fell 37% that year and the cash came from a US$39M inventory drawdown plus the US$49M SBC non-cash add-back. That is working-capital release during a contraction, not durable cash generation. By FY2023 CFO was back to −US$15M, and the working-capital lifeline has not recurred.

There is no factoring, no supplier finance, no recourse receivable arrangement, no bill-and-hold and no acquisition working-capital trick in the disclosures. The cash flow statement itself is a clean reconciliation from net loss through depreciation, amortization, share-based compensation, impairment charges and changes in working capital [18]. The most aggressive cash-flow item is the use of short-term investments as a cash management buffer — FY2025 investing CF was +US$35M because Yatsen sold US$127M of short-term financial products to fund operations and buybacks while purchasing US$86M of fresh ones. This is mechanical, transparent, and correct under US GAAP, but it means investing-line CF in the FY2024-25 window includes ~US$663M of inter-bank cash rotation that visually flatters the bottom line of the cash-flow statement.

Free cash flow has been negative every year since FY2018 with the single exception of FY2022. FY2025 FCF was −US$20M against −US$12M net loss; the US$8M gap is capex (US$6M) plus the inventory build (US$18M). Acquisition-adjusted FCF (FCF minus acquisitions) is identical to FCF because there were no acquisitions in FY2023-25.

Metric Hygiene

Management has been conservative on non-GAAP for a still-loss-making business. The headline non-GAAP EPADS Yatsen reports excludes share-based compensation and intangibles amortisation, and Q4 FY2025 was +US$0.07 versus GAAP loss — but Yatsen consistently reconciles the non-GAAP figures in earnings releases and the gap is shrinking because both adjustments are shrinking organically. Segment reporting (Color Cosmetics Brands, Skincare Brands, others) reconciles segment net revenues and income (loss) from operations to consolidated totals with an unallocated line [19].

No Results

Two non-GAAP-adjacent observations. First, SBC of US$291M in FY2020 (36% of revenue) is the dominant non-cash item in the IPO-year loss and reflects vesting of pre-IPO option grants triggered by listing. By FY2025 SBC has compressed to US$8M (1.4% of revenue), well within normal for a public consumer company. Second, Yatsen has spent US$196M on share buybacks while losing US$332M of net income across FY2022-25, retiring approximately 26% of total ordinary shares since the IPO. Buying back stock during a loss is reasonable if the stock is undervalued and the balance sheet is funded, both of which apply, but it converts US$196M of cash equity into a smaller share count rather than cushioning the operating cash burn — which is one of the reasons cash went from US$876M (FY2020) to US$144M (FY2025). The new March 2026 US$120M convertible notes deal partly reverses this with new dilutive paper.

What to Underwrite Next

Five concrete items for the FY2026 20-F or any pre-print 6-K disclosures:

Related-party purchases over US$53M (FY2025). Track whether the line grows again, whether the counterparties move from "companies over which we exercise significant influence" toward "company controlled by our chief executive officer" (a distinction the disclosure already makes — see [20]), and whether the joint-venture financial guarantee is called [21]. Item 7 disclosure is the only place this appears — read it first. The signal that would downgrade the grade to Elevated: counterparty migration to founder-controlled vehicles, or the line crossing 50% of cost of revenue.
DIO and inventory write-down. FY2025 closed at 198 inventory days, with FY2025 inventory build of US$18M cited as the single largest CFO drag. If FY2026 ends above 220 days and gross margin compresses, expect an inventory write-down inside FY2026 cost of revenue. Compare the inventory dollar growth to revenue growth at every quarter.
Goodwill at US$22M. Eve Lom-related goodwill has been written down twice. Galénic and DR.WU goodwill components remain. If Skincare Brands revenue growth decelerates below 20% in FY2026, run the Yatsen impairment test sensitivity on the remaining US$22M — a third impairment would be a stronger pattern than two.
Hillhouse / Trustar / founder US$120M convertible notes. The convertible-notes private placement (RMB-denominated, ~US$120M aggregate in two tranches, with Trustar Capital and CEO Jinfeng Huang) was announced March 11, 2026 [22]; the first tranche closed May 21, 2026 with Hillhouse affiliates joining as co-investors [23]. Watch the second-tranche closing date, the conversion price, the warrants' strike, and whether the founder's economic stake increases enough to alter the dual-class voting balance. Conversion mechanics and any change-of-control trigger language matter.
Operating cash flow inflection. FY2025 CFO of −US$13.5M was essentially equal to net loss — there is now no accrual cushion left. For FY2026 to print positive CFO, either gross margin needs to hold above 78% on revenue growth that exceeds inventory growth, or working capital must release. The clean test: does CFO turn positive without a one-time benefit?

The signal that would upgrade the grade to Clean: two consecutive quarters of positive GAAP operating income, related-party purchases stable in absolute dollars, and DIO back below 150 days. The signal that would downgrade to Elevated: a third Eve Lom impairment, an inventory write-down disclosed in cost of revenue, or an enlarged convertible-notes facility in which the founder takes the majority of the paper.

Bottom line: the accounting risk at Yatsen is a position-sizing limiter, not a thesis breaker. The reported FY2025 inflection is supported by the cash-flow statement only weakly — most of the optical improvement is the absence of two years of goodwill impairments — but the underlying gross-margin and channel disclosures are credible, the receivable signal is clean, and the largest single legal overhang (the IPO class action) has been dismissed with prejudice. A long position should haircut FY2026 GAAP earnings power for the ~US$19-21M of recurring SBC + intangibles amortisation + inventory true-up risk, and cap position size to reflect that one shareholder controls 90.7% of votes and 100% of the equity-incentive plan administration.

References

Generated Jun 9, 2026 15:44 UTC