Oura filed its S-1 on September 3. Within a day, I had a working three-statement model with a quarterly revenue build and a KPI tab. Every historical figure links back to the filing page it came from. This post walks through how I built it and what the model says.
A fresh S-1 is one of the hardest modeling jobs an analyst faces. There is no prior model to roll forward and no analyst models to adapt. The disclosures rarely line up with the template you want: Oura has a September fiscal year, reports cash flows only for fiscal years and nine-month periods, and carries a deemed dividend on preferred stock that makes pre-IPO EPS difficult to compare. The old way was to print 200 pages and key numbers by hand, then spend the next day hunting the typo that broke the balance sheet.
Daloopa changes the first half of that job. The S-1 was captured by Daloopa within a day of filing, tagged at the line-item level. I built the model using Daloopa’s MCP (Model Context Protocol) integration, which pulls their structured financial data directly into the LLM. My job became structuring the model and forming a view, not transcription.
The workflow
The build itself took one working session. Seven tabs: Summary, Assumptions, Income Statement,
Balance Sheet, Cash Flow, KPIs, and Valuation. Blue text on yellow fill is an editable input. Black is a formula within the tab. Green is a link to another tab. Check rows on every statement compare the computed subtotal to the reported figure, and the balance sheet check row is zero in every column. Every data point pulled through the MCP is hyperlinked to its original source for instant auditability, so any number in the model traces back to the S-1 page it came from.
What the filing shows
Oura sells a ring and a subscription. Hardware is still the larger line, but membership is compounding faster and carries the recurring economics. The table below is the core of the model: two fiscal years and the two most recent comparable June quarters, all sourced.
Table 1: Reported fundamentals ($ in thousands unless noted)
| Metric | FY24A | FY25A | FQ3’25A | FQ3’26A |
|---|---|---|---|---|
| Rings sold (millions) | 1.0 | 2.3 | 0.6 | 1.0 |
| Paid members, end of period (millions) | 1.3 | 2.9 | 2.5 | 5.0 |
| Hardware revenue | 331,203 | 749,393 | 190,014 | 315,812 |
| Membership revenue | 75,548 | 158,463 | 44,276 | 92,912 |
| Total revenue | 406,751 | 907,856 | 234,290 | 408,724 |
| Cost of revenue | 142,657 | 436,844 | 114,438 | 165,037 |
| Sales and marketing | 108,410 | 202,217 | 50,299 | 103,208 |
| Adjusted EBITDA | 36,972 | 74,866 | 16,507 | 10,710 |
| Net income (loss) | 3,649 | 12 | 3,007 | (9,966) |
Every figure links to its Daloopa source cell in the S-1. Fiscal year ends September 30. FY25 net income of $12 thousand is as reported and is effectively breakeven.
Revenue more than doubled in FY25 to $907.9 million. The June 2026 quarter grew 74% year over year to $408.7 million, and the December 2025 holiday quarter was $451.0 million on 1.4 million rings sold. Paid members doubled in twelve months to 5.0 million. The membership line alone is now running at a $372 million annual pace
Holiday quarter revenue and rings sold are linked to the S-1. FQ3’26 growth and the annualized membership run rate are computed from the linked quarterly figures: 408,724 / 234,290 and 92,912 x 4.
The other thing the filing shows is a balance sheet that was rebuilt during fiscal 2026. Oura repurchased $1.17 billion of common and preferred stock in the nine months to June, funded from cash and $377 million of new debt, and booked a $985 million deemed dividend against accumulated deficit. Cash fell from $764 million to $372 million and debt went from near zero to $380 million. Operating cash flow over the same nine months was $328 million.
Table 2: Cash, capital and cash generation ($ in thousands)
| Metric | Sep-24 | Sep-25 | Jun-26 |
|---|---|---|---|
| Cash and cash equivalents | 105,200 | 764,157 | 371,764 |
| Total debt outstanding | 138,326 | 3,226 | 380,135 |
| Redeemable convertible preferred stock | 492,770 | 1,594,980 | 1,499,751 |
| Operating cash flow (period) | 28,227 | 121,693 | 328,008 |
| Capital expenditures (period) | (13,189) | (21,724) | (66,240) |
| Repurchase of common and preferred (period) | (37,404) | (308,018) | (1,172,897) |
| Deemed dividend to preferred (period) | (12,200) | (186,100) | (985,023) |
Period columns for Sep-24 and Sep-25 are full fiscal years; the Jun-26 column is the nine months ended June 30, 2026. Balance
sheet items are as of the stated date.
How the model is built
Revenue is a unit build, not a growth rate
Hardware revenue is rings sold times ASP. Membership revenue is average paid members times
quarterly ARPU. Both drivers are backed out of the reported quarters so the historical ASP and ARPU are visible, then projected. Seasonality is handled by growing each quarter off the same quarter a year ago rather than sequentially, which keeps the December quarter as the peak. Rings sold grow 40% year over year in the September 2026 stub, 30% through fiscal 2027, and 20% in fiscal 2028. ASP rises 3%, then 2%. Net member adds are set at 0.55 per ring sold, between the 0.70 ratio of fiscal 2025 and the 0.40 ratio of the June 2026 quarter. Quarterly ARPU steps from $19.50 to $20.50.
Costs are ratios, working capital is days
Cost of revenue and each operating expense line are percentages of revenue, with modest leverage over time: cost of revenue from 46% to 44%, sales and marketing from 24% to 22%, R&D from 20% to 17%, and G&A from 12% to 9%. Working capital defaults reference the June 2026 balance sheet against
trailing-twelve-month revenue and cost of revenue. Debt is held flat at $380 million with a 7% coupon. Cash earns 3.5%. Positive pretax income is taxed at 25%.
Cost of revenue is held near the fiscal 2025 level rather than extrapolated from the 57% to 60% blended gross margin of the last two quarters. The reason is channel mix. Oura started as a direct-to-consumer business and now sells through wholesale partners, including Costco, Amazon, Best Buy, and Target, across 8,400 retail doors. Wholesale revenue went from $110.1 million in fiscal 2024 to $353.3 million in fiscal 2025 and $490.7 million in the nine months to June 2026, about 49% of hardware revenue. It grew 77% year over year in that nine-month period, against 72% for direct. The retailer keeps its margin, and revenue per ring has drifted from $332 in fiscal 2024 to $326 in fiscal 2025 and $311 in the nine months to June 2026 while list prices held. Oura says it expects reliance on retail partners to increase and names channel mix as one of the four drivers of hardware gross margin. Manufacturing scale on Ring 4 pushes the other way, so the model nets the two at a 44% to 46% cost of revenue rather than assuming the June quarter margin holds.
The stub quarter problem
The S-1 gives seven quarters of income statement data but only fiscal-year and nine-month cash flows. The model handles this by treating FY26E as nine months actual plus a September quarter stub. The September 2026 balance sheet rolls forward from the June 2026 actual using that stub, then the model goes annual. D&A and share-based compensation are disclosed only in the cash flow statement, so the FY26E totals are inputs and the stub is the plug. The KPIs tab derives FQ4’25 cash flow as FY25 less nine months FY25, which is the only way to get a quarterly free cash flow figure out of this filing.
The KPIs the filing does not give you
Oura discloses one retention statistic: 85% weighted-average 12-month paid-member retention as of
June 30, 2026, given by fiscal-year vintage as roughly 81% for fiscal 2023 cohorts, 85% for fiscal 2024, and 87% for cohorts started in the nine months to June 2025. The figure includes winbacks within the 12-month window and members in a 28-day payment grace period. It does not disclose gross adds, churn, membership gross margin, or cohort revenue. The KPIs tab builds proxies for the metrics an investor wants and shows how each is constructed so a reader can disagree with the method rather than the number.
- Gross member adds = net adds plus churn, where churn is the opening base times one minus the
quarterly retention rate (85% to the power of one quarter). - CAC per gross member add = sales and marketing expense divided by gross adds.
- CAC payback = CAC per gross add divided by monthly gross profit per member, using blended
gross margin because membership margin is not broken out. - Membership NRR proxy = annual retention times one plus year-over-year ARPU change. It
excludes hardware repurchases.
Table 3: Derived unit economics
| Metric | FY25A | FQ1’26A | FQ2’26A | FQ3’26A |
|---|---|---|---|---|
| Net member adds (millions) | 1.60 | 0.50 | 1.20 | 0.40 |
| Gross member adds (millions) | 1.80 | 0.62 | 1.34 | 0.58 |
| S&M per ring sold ($) | 88 | 58 | 91 | 103 |
| CAC per gross member add ($) | 113 | 133 | 55 | 177 |
| Membership ARPU, monthly ($) | 6.29 | 6.89 | 6.88 | 6.45 |
| Blended gross margin | 51.9% | 47.8% | 57.2% | 59.6% |
| CAC payback (months) | 34.5 | 40.3 | 13.9 | 46.0 |
| Membership NRR proxy | 110% | 88% | 87% | 86% |
Three things stand out. First, the quarterly CAC figures swing widely because net adds are lumpy: 1.2 million in the March quarter and 0.4 million in June. That pattern reflects rings sold in the holiday quarter converting to paid members after the 30-day trial, rather than a change in acquisition efficiency. Annual figures are the better read, and on that basis Oura spends roughly $110 to $140 to acquire a paying member. Second, payback measured against membership gross profit alone is long, around three years. That measure excludes the gross profit on the ring itself, which the same marketing spend also generates. Third, the NRR proxy sits below 100% in every recent quarter. With 15% annual churn, membership revenue only compounds if ARPU rises or the ring base keeps growing. Right now, the ring base is doing all the work.
What the model says
With the assumptions above, the model produces the following fiscal year path. FY26E is nine months actual plus the September stub, so it is mostly reported.
Table 4: Model outputs ($ in thousands unless noted)
| Metric | FY25A | FY26E | FY27E | FY28E |
|---|---|---|---|---|
| Total revenue | 907,856 | 1,540,694 | 2,161,271 | 2,789,074 |
| YoY growth | 123% | 70% | 40% | 29% |
| Gross margin | 51.9% | 54.4% | 55.0% | 56.0% |
| Adjusted EBITDA | 74,866 | 114,164 | 222,611 | 398,838 |
| Adjusted EBITDA margin | 8.2% | 7.4% | 10.3% | 14.3% |
| Net income | 12 | 50,845 | 55,019 | 163,737 |
| Free cash flow | 99,969 | 276,196 | 236,653 | 376,041 |
| Ending cash, including restricted | 860,546 | 386,192 | 622,845 | 998,886 |
| Paid members, end of period (millions) | 2.9 | 5.4 | 8.2 | 11.5 |
FY25A revenue, Adjusted EBITDA, net income and paid members are linked to the S-1. FY25A free cash flow is reported operating cash flow (source 179390933) less reported capex (source 179390935). FY25A ending cash is cash of 764,157 plus restricted cash of 96,389 (sources 179390851 and 179390852). Projections are model output and recalculate from the yellow input cells.
Fiscal 2026 lands near $1.54 billion of revenue, up 70%, with $276 million of free cash flow. The model is more conservative than the recent print on margin: fiscal 2026 Adjusted EBITDA margin dips to 7.4% because the June quarter carried a step-up in R&D and G&A ahead of the IPO, and I let those ratios only partly normalize. Below the line, net income barely moves from fiscal 2026 to fiscal 2027 because the 25% tax rate applies to a full year of positive pretax income, and the 7% coupon on $380 million of debt runs against a smaller cash balance. Free cash flow dips in fiscal 2027 for the same reason, plus working capital scaling with revenue off the June 2026 days. By fiscal 2028, the model has 11.5 million paid members, $2.8 billion of revenue, and a 14% Adjusted EBITDA margin. Cash rebuilds to $1.0 billion before any IPO proceeds
The inputs that matter most
A model is only useful if you know which cells matter most. Four of them are key.
- Rings sold growth. The 30% assumption for fiscal 2027 sits well below the 67% unit growth in the
June quarter. It is the single largest driver of every downstream number. Ten points of unit growth
is more than $120 million of fiscal 2027 revenue. - Retention. The 85% figure is a weighted average across vintages that have improved from 81% to
87%, so the blended number should rise as the larger recent cohorts dominate. The softer spot is
the definition, which counts winbacks and grace period members as retained. If a stricter measure
puts it at 80%, annual churn rises by a third, the gross adds required to hit the same net member
count rise with it and CAC payback stretches accordingly. - Membership gross margin. The mix shift to membership is what lifts blended gross margin in the
model. The filing does not break the margin out. If membership margin is closer to hardware
margin than I assume, the fiscal 2028 EBITDA margin output is too high. - Channel mix. Wholesale was 27% of revenue in fiscal 2024 and 40% in the first nine months of
fiscal 2026, and Oura expects its reliance on retail partners to grow. Every ring that moves from
direct to Costco, Amazon, Best Buy or Target hands a slice of hardware gross profit to the retailer,
which is already visible in revenue per ring. The model assumes cost of revenue improves only
modestly to 44% for this reason. If wholesale keeps outgrowing direct, hardware margin falls
further and the blended margin gains from membership mix are spent offsetting it rather than
expanding EBITDA. Wholesale shares are computed from the linked channel revenue above
(110,105 / 406,751 and 490,715 / 1,214,506).
The bull case is that Oura is a subscription business with a hardware on-ramp: membership revenue
grew 110% in the June quarter, and the installed base of members is the annuity. The bear case is that the growth is still a hardware cycle; that a 15% churn rate requires continuously replacing part of the member base; that the shift to wholesale keeps moving hardware margin to retailers; and that competition on the ring form factor compresses ASP before the membership base is large enough to carry the P&L. The model lets you test both without rebuilding anything. Change the yellow input cells and every statement, KPI, and multiple recalculates. Daloopa’s MCP did the heavy lifting on retrieval and sourcing, which left time for the channel-mix and retention work above.
Data sourced from Daloopa. This is research commentary, not investment advice.
