Revenue has stabilised. Margins have not.
Nike lost $5.0B of revenue from its FY2024 peak and four points of operating margin. The analysis shows a structural break in FY2025, with margin recovery, not demand, as the main lever to FY2029.
Revenue and operating margin
Bars: revenue ($B). Line: operating margin (right axis).
Where the $5B went
Change in revenue FY2024 → FY2026 by segment ($B)
Operating margin bridge, FY2024 → FY2026
Overhead that did not shrink with revenue explains about 60% of the margin loss.
A regime break in FY2025
Regression on 80 region-quarters: log revenue with region trends, quarter seasonality and a FY2025+ shift. R² ; the break is significant (F-test p < 0.001).
Actual quarterly revenue ($B) vs model fit (dashed)
Estimated effects
95% confidence intervals, robust (HC3) standard errors
| Region | Trend p.a. | FY25+ shift | p (shift) |
|---|
Before FY2025 North America and APLA compounded at ~6.5% a year. Since then, each sits 13–17% below that path.
Regional revenue, FY2017–FY2026
$B, stacked
A flat outlook with honest uncertainty
Five models plus an ensemble, compared on rolling-origin validation. Bands are 80% and 95% prediction intervals calibrated to real out-of-sample error.
Model comparison
Out-of-sample MASE (below 1 beats a seasonal-naive forecast). Selected model in bold.
| Model | MAPE | MASE |
|---|
Who is gaining and losing ground
K-means clustering (k = 3) of each region-year on growth, two-year growth, growth versus the company, and share change. Segments are relative to company performance.
Segment by region and year
Cell value = year-on-year revenue growth
Segments in principal-component space
Two components capture 91% of the feature variance
Segment profiles
Average of member region-years
| Segment | n | YoY growth | vs company | Share chg |
|---|
In FY2026 North America is the only region gaining ground. Greater China and Converse are both losing it.
Moving together, and apart
Correlations on revenue levels are inflated by shared trend. Growth-rate correlations show the real co-movement and the cannibalisation signals.
Regional correlation
Greater China is decoupled (r = 0.03 with North America): it diversifies revenue but will not recover alongside the West.
Cannibalisation tests
Correlation of annual growth rates, n = 9
| Pair | r | Verdict |
|---|
Nike Direct vs wholesale
During the DTC push (FY19–23) Direct added $9.6B while wholesale still grew $2.0B: gains were additive, not taken from wholesale.
Scenario planner
Start from a preset, then move any driver. The P&L rebuilds from FY2026 actuals. Bear and Bull are calibrated to the P10 and P90 of a 20,000-run Monte Carlo.
Revenue path
Presets (thin) and your scenario (orange), $B
| FY2027 | FY2028 | FY2029 |
|---|
Drivers
Growth rates are applied each year, FY2027–FY2029
What moves FY2029 EBIT
Change vs Base ($B) when one driver moves to Bear / Bull
FY2029 EBIT distribution
Monte Carlo, 20,000 runs. Probability of beating the FY24 revenue peak: