Key points
- Payrolls were weak and heavily revised. The BLS said nonfarm payrolls rose just +73,000 in July, and May–June were revised down a combined −258,000—a big credibility test for headline figures.
- Two service-sector gauges told opposite stories. ISM Services PMI sat at 50.1 (near stall, with employment 46.4), while S&P Global’s Services PMI jumped to 55.7 (strong growth).
- Private payroll tracker diverged too. ADP estimated +104,000 private jobs in July, further widening the gap with the BLS print.
- Politics raised the temperature. After the July report, BLS Commissioner Erika McEntarfer was dismissed, sparking warnings about interference and the integrity of federal statistics.
The spark: a weak July and big downward revisions
On Friday, August 1, 2025, the Bureau of Labor Statistics reported that payrolls “changed little” in July (+73,000) and the unemployment rate held at 4.2%. The same release revealed larger-than-normal revisions: May was cut from +144,000 to +19,000, and June from +147,000 to +14,000—−258,000 combined. That’s not manipulation; it’s the BLS doing what it’s supposed to do: incorporate late survey responses and updated seasonal factors. But it did fuel the current debate over reliability.
The composition looked familiar: health care (+55,000) and social assistance (+18,000) carried the gains, while the federal government (−12,000) and several major service categories showed little change. Average hourly earnings rose 0.3% m/m and 3.9% y/y.
Two PMIs, two stories
The Institute for Supply Management’s Services PMI for July printed 50.1, barely above the 50 growth/contraction line. Under the hood, employment contracted (46.4) and prices ran hot, a pattern consistent with firms delaying backfills while grappling with rising costs. ISM respondents also highlighted tariff pressures and supply friction.
By contrast, S&P Global’s US Services PMI jumped to 55.7, its strongest of 2025, pointing to robust output growth on domestic demand. When two respected surveys disagree, it doesn’t mean one is “wrong”—they sample different firm sets and ask different questions. Divergences are common around turning points.
ADP vs. BLS: why the gap?
Payroll processor ADP estimated +104,000 private-sector jobs in July, while the BLS establishment survey showed +83,000 for total private and +73,000 overall including government. Differences stem from coverage (BLS includes government; ADP is private-only), methods (model vs. survey frame), and timing. Month to month, these series can wander; over longer spans, they usually rhyme.
Hiring Plans on Hold? The Data Says You’re Not Alone
With payroll growth slowing and past gains revised sharply lower, finding the right talent fast has never been more critical.
Post a Job Now →Politics enters the chat
The underlying anxiety isn’t just statistical—it’s political. Within days of the report, BLS Commissioner Erika McEntarfer was fired, after the White House publicly questioned the agency’s numbers. Economists warned that politicizing revisions risks eroding confidence in an institution whose methodologies are deliberately transparent and, yes, designed to revise as better data arrive.
Revisions are a feature, not a bug
BLS explains plainly why its payroll figures move: each month, initial estimates get two standard revisions as more employer responses come in and seasonal factors are recalculated. Once a year, a benchmark revision re-anchors levels to unemployment-insurance tax records (QCEW)—the closest thing we have to a near-census of payroll jobs. In short: first prints are provisional by design.
It also matters that the US has two labor surveys: the household survey (CPS) that yields the unemployment rate and the establishment survey (CES) that produces payroll counts. They cover different universes—self-employed people appear in CPS but not CES; multiple-job holders are counted more than once in CES but once in CPS—so they will diverge at times.
How operators should read the numbers—without getting whiplash
1) Triangulate, don’t fixate. Pair the BLS payroll trend with PMIs and private trackers. When PMIs split (ISM ≈ 50, S&P ≈ 56), weight them with your exposure: if you sell to big corporates and public entities, ISM’s mix may feel closer; if your book is mid-market/consumer-facing, S&P’s strength may be your reality.
2) Focus on direction + revisions. The three-month average and the revision pattern tell you more than one noisy print. July’s downward revisions argue for caution on hiring plans even if your order book looks decent.
3) Watch the service’s price gauges. ISM’s Prices index spiked again, a classic margin squeeze warning for service businesses (software, logistics, pro services, hospitality). Build 5–10% cost contingencies into Q4 budgets until this cools.
4) Separate cyclical from structural. The health-care and social-assistance engine remains resilient—don’t starve growth roles there—while temp help, business support, and some federal categories are soft. Tilt your pipeline accordingly.
5) Protect your own signal. Maintain leading internal indicators—fill rates, time-to-accept, offer/accept ratios, pipeline velocity—so you’re not over-reacting to every headline.
What to watch next (with dates)
- Next ISM Services: early September per ISM’s release calendar; continued sub-50 employment would confirm softening hiring appetite.
- Next jobs report: Friday, Sept 5, 2025 (August data). Expect markets to parse revisions first.
- ADP: Sept 4, 2025 at 8:15 a.m. ET—an earlier read-through on private hiring.
Bottom line
We’re not witnessing “fake data”; we’re seeing noisy data in a slowing, policy-shocked economy. The best protection for hiring managers and founders is to triangulate: read BLS prints (and their revisions), cross-check with ISM/S&P, and sanity-check with private trackers like ADP. And resist the urge to politicize routine revisions. The statistical system is built to admit new information—that’s how it stays honest.
FAQs
Why did the July jobs numbers get revised so much?
Because more employer reports arrived and seasonal factors were recalculated—standard practice that improves accuracy after the first estimate. July’s −258,000 two-month revision was larger than usual, but still part of this normal process.
Which services PMI should I trust—ISM or S&P Global?
Both are credible; they survey different panels and use different methods. Short-term divergences aren’t rare. Look at direction, not just levels, and combine them with actual orders and hiring plans in your sector.
Why does ADP often differ from the BLS?
ADP covers the private sector via payroll data and applies its own modeling; BLS surveys businesses and governments and revises as new data arrive. Month-to-month, gaps happen; over time, trends converge.
