The August Flash PMI Is Friday. The Numbers That Matter Are Buried Three Rows Down

by Alina Castellanos · August 16, 2026

Economics Correspondent, US Market Current

The next scheduled read on the American economy is a survey, not a government statistic, and it arrives on Friday. S&P Global's flash US Composite PMI for August is the first hard-ish evidence on the current month that anyone will see, and the parts of it that will move the argument about the economy are not the headline numbers most wire services will lead with. They are the input-price and output-price sub-indices, and in July those two lines told a story sharply at odds with the cheerful top-line print above them.

Start with what the July surveys actually said, because the flash and the final told noticeably different versions of it. S&P Global's flash release for July, embargoed until 9:45 a.m. Eastern on Friday, July 24 and built from responses collected between July 9 and July 23, put the Flash US Composite PMI Output Index at 53.6, up from 51.9 in June. S&P Global described that as an eight-month high and the fastest growth since November 2025. The flash Services Business Activity Index was 53.6 against 51.2 in June. The flash Manufacturing PMI was 53.8 against 53.9, and the flash Manufacturing Output Index was 53.6 against 56.2 — both flagged by S&P Global as four-month lows, and a sizeable step down in the pace of factory output growth even as the composite jumped.

S&P Global's own strapline on that release did not lead with growth alone. It read that business activity growth accelerated to an eight-month high in July, but that selling prices rose at the fastest rate for nearly four years. That second clause is the story. At the composite level, on those preliminary numbers, S&P Global said the overall rise in charges was the steepest since August 2022, and that input cost inflation had risen to its highest since May 2025 — cooler manufacturing cost growth, in the firm's phrasing, more than offset by a 14-month high in services. Services charge inflation climbed to the highest in just under four years, while factory-gate price inflation remained sharp, albeit reduced.

Then the final data arrived and moved. S&P Global published final July manufacturing on Monday, August 3 and final July services and composite indicators on Wednesday, August 5. The final Services Business Activity Index came in at 54.6 against June's 51.2, a full point above the 53.6 flash. The final Composite Output Index was 54.5 against June's 51.9, nearly a point above the 53.6 flash. Final manufacturing barely moved, printing 53.9, unchanged from June's 53.9 and a tenth above its own flash.

The revision changed the price story too, and not in the direction the flash had implied. On the preliminary numbers S&P Global called composite input cost inflation the highest since May 2025 and the overall rise in charges the steepest since August 2022. On the final numbers the composite input-price line was described as the highest since November 2022, while composite selling prices were said to have risen at the quickest pace in exactly one year. A hotter cost reading and a cooler prices-charged reading than the flash had produced, from the same panel in the same month. The composite output index moved on its label as well: an eight-month high on flash data, described in the final release as the strongest since October 2025. Anyone still quoting the August 2022 selling-price line is quoting a number that S&P Global's own completed sample superseded eleven days later.

That revision is worth dwelling on, because it is the single most useful thing to know before Friday. The flash is not a preview of the final; it is a different number computed from a smaller sample. S&P Global's own methodology note on the release states that flash data are calculated from around 80 to 90 percent of total responses and are intended to provide an accurate early indication of the final data. That is the firm's stated coverage range, and it is a range, not a point.

Revisions in the other direction happen too, and recently. S&P Global's June flash, collected June 11 to 22, reported a Flash US Manufacturing PMI of 55.7 and a Flash Manufacturing Output Index of 57.7. By the time the July releases carried June as their comparison month, June manufacturing stood at 53.9 and June manufacturing output at 56.2. The composite told a smaller version of the same story: 52.2 in the June flash, 51.9 in the comparison column of the July releases. In May, S&P Global's flash put the composite at 51.7, services at 50.9 and manufacturing at 55.3; the June flash's comparison column showed May at 51.5, 50.7 and 55.1 respectively. Three consecutive months, three sets of movements between flash and final, and in June's case a manufacturing revision of nearly two points.

The second thing to hold onto is what a diffusion index is. A PMI reading is constructed from the share of surveyed purchasing managers reporting improvement versus deterioration against the prior month, seasonally adjusted, with 50 the no-change line. It measures direction and breadth — how many firms saw something get better or worse — not magnitude. A reading above 50 that falls from the month before still describes expansion, only a less widespread one. A price sub-index at a four-year high says that an unusually broad share of the panel reported paying or charging more than they did the month before. It does not say by how much, and it cannot be converted into a percentage change in the consumer price index. Anyone who tells you Friday's price line implies a specific CPI print is doing something the survey does not support.

That distinction matters more than usual right now because the July surveys contained a genuine internal disagreement, and the August flash is the next chance to see which side of it holds. On the services side, the final July release described the steepest rise in new work since November 2025, a composite new business reading S&P Global called a 19-month high, and services job creation at its strongest in eight months, with private sector employment rising for the first time since April. Business confidence about the coming 12 months was the strongest since November 2025.

On the manufacturing side, the same month looked much worse, and S&P Global said so in its own headline: sector expansion holding steady but masking softer production and sales growth. The final July manufacturing release described production volumes rising at the weakest pace since March, the expansion in new order intakes easing for a third successive month, and staffing rising only marginally, with anecdotal evidence suggesting vacancies had been broadly filled. Input cost inflation moderated to a four-month low but remained stronger than the series average, and firms raised their own charges where possible, at a similarly weaker pace. Chris Williamson, S&P Global Market Intelligence's chief business economist, argued in that release that the steady headline concealed warning signs about the future growth trajectory: production rose at a markedly slower rate, he said, on a third month of weakened growth in new business, itself a reflection of reduced inventory building after the especially strong precautionary stock accumulation of the second quarter. He pointed to increased supply chain delays, falling exports and further customer pushback on high prices, tied the elevated cost pressure to high energy prices and tariffs, and said manufacturers' optimism about growth prospects had slipped to its lowest since October 2025.

Williamson's read on the services-led acceleration came with its own set of qualifiers, and they are the reason Friday matters rather than a reason to discount it. He opened the final release by noting that July had come in stronger than the earlier flash estimate. He then attributed a meaningful share of the improvement to temporary factors: the largest gains came from consumer-facing service providers, where spending surged at a rate not seen for over four years, which he linked to the FIFA World Cup and US Independence Day events. He also noted that businesses had benefited in early July from a tailwind of reduced geopolitical uncertainty and lower oil prices, and warned that with hostilities in the Gulf escalating as the month progressed, the geopolitical environment was likely once again acting more as a headwind to growth while exacerbating already-elevated price pressures. Both of those tailwinds were, by construction, absent from most of the August survey window.

The GDP signal the survey generates shifted with the revision as well, which is a useful reminder that these translations are elastic. In the July flash, S&P Global said the data were broadly consistent with GDP growing at an annualized 2.0 percent, against a 1.2 percent pace signalled for the second quarter. In the final release twelve days later, with the higher composite reading, Williamson put the implied figure at 2.3 percent annualized following a 1.5 percent increase indicated for the second quarter. Same survey, same month, same methodology, different sample completeness, and three tenths of a percentage point of implied growth in between.

One more distinction, because it is routinely blurred. The S&P Global PMI is not the ISM survey. They are separate monthly surveys of separate panels with different sector weights, different question sets and different seasonal adjustment, and they have disagreed repeatedly through this cycle. The Institute for Supply Management's reports on business are published on their own schedule, typically in the first business days of the following month, and its prices index is not the same series as S&P Global's input-price sub-index. ISM's July reports were not accessible for this piece, so no ISM figure is cited anywhere in this article; every number above and below comes from S&P Global.

As for timing: FXStreet's economic calendar lists the next US S&P Global composite PMI release for August 21 at 13:45 UTC, which is 9:45 a.m. Eastern, consistent with the embargo time stamped on S&P Global's own recent flash releases. August 21 is a Friday. The calendar lists no consensus forecast for the release and carries 54.5 — the final July composite — as the previous value. We were not able to load S&P Global's own US release calendar to confirm the date from the primary source, so treat it as a calendar-provider date rather than a publisher-confirmed one.

It is worth correcting a piece of scheduling folklore that has been circulating this week. The August flash does not land during the Kansas City Fed's Jackson Hole symposium. The Kansas City Fed's own symposium page gives the 2026 dates as August 27 to 29, with the theme "Financial Innovation: Implications for Payments and Policy." That is the following week — six days after the flash, not the same day. The two events are both on the August calendar and both will be read through the same lens, but the flash will have been digested, argued over and half-forgotten by the time anyone arrives in Wyoming.

So what would actually be informative on Friday? Three things, in descending order of usefulness. First, whether the composite output-price sub-index stays near the one-year extreme the final July data described or backs off — and note that the flash's more dramatic August 2022 framing is exactly the sort of preliminary characterisation the completed sample walked back, so Friday's price headline should be held loosely for two weeks. A second consecutive month of unusually broad selling-price increases is a materially different piece of evidence than a one-month spike tied to a World Cup summer. Second, whether the manufacturing and services panels re-converge or diverge further, since a composite built from a decelerating factory sector and a surging consumer-services sector is an average of two unlike things. Third, whether employment holds the ground it recovered in July, when private sector staffing rose for the first time since April.

And the fourth thing, which is really a caution rather than a datapoint: whatever prints Friday, the July releases already demonstrated that S&P Global's own flash can be off by nearly a point on the composite and by close to two points on manufacturing, in either direction — and that the comparison periods attached to the price lines can move as well. The final August manufacturing and services releases will follow in the first business days of September on the pattern the firm has used all year. Between now and then, the flash is an early indication with a stated 80-to-90-percent response base, described as such by the people who produce it, and it should be read that way.