Every GDP print you have ever seen was a draft. The Bureau of Economic Analysis releases each quarter's growth estimate three times — advance, second, third — and then rewrites history in annual updates that can move years of growth. Commentators quote the advance estimate because it is first; the honest readers wait, because the differences between vintages are often the story.
What the number actually measures
GDP is the value of final goods and services produced in a period, expressed after inflation as real growth at a seasonally adjusted annual rate. Its identity is consumption plus investment plus government spending plus net exports. Two features trip readers immediately. Annualized rates multiply one quarter's growth by roughly four — a 2% annualized print is a 0.5% quarterly move. And net exports subtract: imports enter negatively, so a surge of import buying — like the inventory front-running that preceded tariff deadlines in 2025 — mechanically dents the quarter regardless of domestic demand.
Why the revisions are so large
Because the early vintages run on partial data. The advance estimate arrives about a month after quarter-end with roughly incomplete source data; quarterly and annual surveys, tax records, and census fills arrive over months and years. Typical revisions are a few tenths of a point, but turns in the cycle are where revisions bite: recession quarters have been revised from positive to negative, and inventory and trade components — the volatile ones, and precisely the ones tariff-era distortions moved — carry the widest revision bands. The annual update in late September re-benchmarks everything; the level of GDP itself gets redefined every few years as the statistical framework updates.
What should a reader track instead of the headline?
Final sales to domestic purchasers — GDP minus inventories minus net exports — which answers the question people actually ask: is domestic demand growing? The two-year average of growth, which smooths the pandemic-era distortions that still echo in comparisons. And the composition: growth led by business equipment investment means something different from growth led by federal outlays or inventory swings. For tariff-era reading, the trade and inventory lines deserve separate inspection, because policy timing — companies importing ahead of effective dates, then drawing stocks down — swings quarters without changing the underlying trend.
How does 2026's policy mix show up in GDP?
Through the same components as always. Tariff actions move import prices and volumes, which moves the net-export and consumption-deflator lines — note that tariffs raise the price level of imported goods, a level shift that shows in inflation data rather than a continuing GDP drag. Federal industrial programs — semiconductor incentives, defense ramp — appear in government purchases and business structures investment with multi-quarter lags. Interest costs on federal debt, which crossed a trillion dollars annually as a running expense, crowd the fiscal channel that backstops demand. None of these make the advance estimate wrong; they make it provisional, which is the same thing.
A reading routine
Note the vintage you are quoting and the two revisions behind it. Check final sales to domestic purchasers before celebrating or mourning the headline. Separate inventories and trade from the demand signal. Compare against the same quarter's year-ago level, not the annualized quarter-over-quarter arithmetic. And hold conclusions loosely through the September benchmark. GDP is the best-constructed scoreboard the economy has — and it is still a scoreboard someone is still repainting.
For more context, read Interest now outruns most of the budget.
For more context, read tariffs and consumer prices.
For more context, read Jobs day rewards the prepared reader.