What to Watch

What to Watch

The six prior chapters resolve into two coherent readings of the same company, and the evidence does not yet decide between them: a cheap, cash-backed premiumization story approaching self-funding profit, or a structurally loss-making marketing model slowly spending its IPO windfall. Both are built from the same filings. What separates them is chiefly the marketing bill and a few line items that will move over the next several reports. This chapter reconciles the two and sets out what to watch.

Two readings of the same numbers

The constructive read has real support. Gross margin reached 78.2% in 2025, up from 77.1%, as higher-margin skincare took over the mix [1]. Skincare became the larger segment for the first time, at 53.0% of revenue against color cosmetics' 46.7% [2], and it grew 58.5% again in the first quarter of 2026 as group revenue rose 22.5% [3]. The company posted its first full-year non-GAAP net income (¥8.4 million) and its first quarterly GAAP net profit (¥3.0 million in the seasonal fourth quarter) in 2025 [4]. Against roughly $145 million of net cash and no debt (Valuation and Cash), the market pays almost nothing for the operating brands.

The skeptical read uses the same statements. The 2025 non-GAAP profit is a 0.2% margin resting on roughly ¥100 million of non-cash add-backs; on a GAAP basis the operating line still lost ¥185.8 million (−4.3% of revenue) [5]. Selling and marketing ran ¥2.85 billion, 66.3% of revenue, and has sat in a narrow band near two-thirds of sales for six years through the crash, the category pivot, and hundreds of store closures (Marketing Economics) [6]. The moment the company paid to grow in the first quarter of 2026, the operating loss more than doubled to ¥99.0 million (−9.7% of revenue) [7]. Full-year operations burned ¥94.7 million, and even the profitable fourth quarter still used cash — though not every quarter did: the first quarter of 2025 was cash-generative [8].

Neither read is a strawman. The disagreement is not about the facts but about whether the marketing ratio is a fixed cost of a rented audience or a number that premiumization will bring down with a lag.

Where the two readings collide

Each row below is a fact both sides accept, read two ways, with the evidence that would settle it.

Sources: FY2025 20-F MD&A and notes [9][10]; Q4 FY2025 and Q1 FY2026 results [11][12].

Four of the five rows resolve through the same channel: the trajectory of the marketing ratio and the operating cash it frees or consumes. The case is most sensitive to that one line, which is why the watch-list below is built around it rather than around revenue growth, where the company has already delivered.

What today's price implies

At $3.43 on 28 July 2026, the market marks Yatsen's operating business at roughly a sixth to a third of one year's sales once the cash and securities are stripped out — a fraction of the ~1.9x EV/sales a profitable domestic peer commands, and a price consistent with only a ~2–3% normalized operating margin (Valuation and Cash). The gap between the two readings is therefore the gap between a business that stays near breakeven and one that converts its 78% gross margin into a peer-like operating margin. The chart frames that span against where the company actually sits today and where the peer sits.

Sources: FY2025 and Q1 FY2026 operating margin as reported [13][14]; "priced-in" and "mix lowers marketing" are illustrative scenarios, not forecasts; Proya per competitor financials, as reported (Marketing Economics).

The marketing line is the whole story: at 66.3% of revenue it is both the ~22-point operating-margin gap to profitable peer Proya and the ~¥900m swing between value trap and re-rating on ~¥5.5bn of sales, while Yatsen spends about 21 times as much on marketing as on R&D — so the step-down the upside depends on has no product moat visibly behind it. On roughly ¥5.5 billion of revenue — near the single-analyst 2026–2027 consensus — the distance between the −5% margin of a still-heavy marketing model and the +12% of a mix that pulls S&M toward the high-50s is about ¥900 million of operating profit, which is why the same balance sheet supports both a value-trap read and a multiple-of-today read. Management states in the 20-F that it "intend[s] to continue optimizing this ratio by reallocating marketing spending toward higher-return channels" and "capturing operating leverage across our fixed expenses" [15]; it has not attached a number or a date to that intent, and guides revenue only.

The watch-list

Each item names the line, the filing it appears in, and the threshold that would move the read from one column of the tension table to the other. The nearest read is close: the second-quarter results are scheduled for 20 August 2026.

Sources: guidance and cash-flow figures per Q1 FY2026 and FY2025 results [16][17]; convertible and restricted-assets terms per FY2025 20-F notes [18][19]; next-earnings date per company calendar.

The market pays almost nothing for Yatsen's brands — about 74% of its ~US$322m value is cash and securities, implying a 0.14-0.29x sales operating enterprise value — yet that floor is shrinking as operations burn ¥94.7m a year and is now outranked by US$120m of insider convertible notes whose year-three cash put can withdraw roughly the whole pre-deal net-cash pile (Valuation and Cash). That convertible carries two distinct clocks that are easy to conflate. The First Note has a 364-day maturity that extends to five years only on receipt of the NDRC foreign-debt registration certificate [20]; absent the certificate, a $60 million refinancing question lands within a year. The year-three cash put is exercisable at a 4% internal rate of return (Control and Capital). And ¥721.5 million of net assets remain restricted from upstreaming under PRC law, just under the SEC's 25%-of-equity threshold [21]. None of these is an insolvency trigger for a debt-light company; each is a claim on, or a lock on, the cash floor the cheapness rests on.

Base rates and positioning

Two features of the setup temper how much any single data point should move the read. First, the shares are thinly covered and thinly traded — effectively one analyst carries the consensus (2026 revenue ¥5,238 million, a $4.01 target), so estimate revisions are one modeler's view rather than a market signal, and the stock has drifted from $3.89 in March to $3.43 in late July within a 52-week range of about $2.19 to $3.89. Second, the earnings-surprise record is volatile and has skewed negative: reported results have missed the (sparse) EPS estimates by wide margins in several past quarters. A single better-looking quarter, especially the seasonally strong fourth, should be weighed against that base rate before it is read as the inflection.

Sources: analyst coverage, price history and earnings-surprise record per market data feeds, as reported.

Whether the stabilization proves durable resolves along the watch-list above rather than in any figure already filed: the balance sheet answers the downside, and the marketing line, quarter by quarter, will answer the rest.